Vous êtes sur la page 1sur 187

••• National Ubrary

01 Canada

Acquisitions and
Bibliographie Services Branch
31bllothèque nationale
du Canada

Direction des acqu1s1tior,s et


des services bibliographiques
395 Welhngton Street 395. rue WCninglon
Gnawa.Ontano ûnawa (Onlano)
K1AON4 KlADNO:

NOTICE AVIS

The quality of this microform is La qualité de cette microforme


heavily dependent upon the dépend grandement de la qualité
quality of the original thesis de la thèse soumise au
submitted for microfilming. microfilmage. Nous avons tout
Every effort has been made to fait pour assurer une qualité
ensure the highest quality of supérieure de reproduction.
reproduction possible.

If pages are missing, contact the S'il manque des pages, veuillez
university which granted the communiquer avec "université
degree. qui a conféré le grade.

Sorne pages may have indistinct La qualité d'impression de


print especially if the original certaines pages peut laisser à .
pages were typed with a poor désirer, surtout si les pages
typewriter ribbon or if the originales ont été
university sent us an inferior dactylographiées à l'aide d'un
photocopy. ruban usé ou si "université nous
a fait parvenir une photocopie de
qualité inférieure.

Reproduction in full or in part of La reproduction, même partielle,


this microform is governed by de cette m!croforme est soumise
the Canadian Copyright Act, à la Loi canadienne sur le droit
R.S.C. 1970, c. C-30, and d'auteur, SRC 1970, c. C-30, et
subsequent amendments. ses amend~ments subséql.>ents.

Canada

Modelling and Simulation of
Electric Mining Shovels

by

Hongjin Wu

B.Se. M.Eng.

• Department of Mining and Metallurgiea! Engineering


MeGill University,
MontréaL Québec, Canada

A thesis submitted to the Faeulty of Graduate Studies and Researeh


in partial fulfillment of the requirernents for the degrcc of
Doetor of Philosophy

May 18, 1995


© Hongjin Wu
••• National Ubrary
of Canada
Acquisitions and
Bibliographie services Branch
Bibliothèque nationale
du Canada
Direction des acquisitions et
des services bibliographiques
39S Wellington Slreet 395. rue Welhngton
Ottawa. Onlano On...a (Onlanol
K,A 0N4 K1A QN4

TIIE AUTHOR RAS GRANTED AN L'AUTEUR A ACCORDE UNE LICENCE


IRREVOCABLE NON-EXCLUSIVE IRREVOCABLE ET NON EXCLUSIVE
LICENCE ALLOWING TIIE NATIONAL PERMETTANT A LA BIBLIOTIIEQUE
LIBRARYOF CANADATO NA.TIONALE DU CANADA DE
REPRODUCE, LOAN, DISTRIBUTE OR REPRODUIRE, PRETER, DISTRIBUER
SELL COPIES OF HISfHER TIIESIS BY OU VENDRE DES COPIES DE SA
ANY MEANS AND IN ANY FORM OR TIIESE DE QUELQUE MANIERE ET
FORMAT, MAKING THIS TIIESIS SOUS QUELQUE FORME QUE CE SOIT
AVAILABLE TO INTERESTED POUR METTRE DES EXEMPLAIRES DE
PERSONS. CETTE TIIESE A LA DISPOSITION DES
PERSONNE INTERESSEES.

TIIE AUTHOR RETAINS OWNERSHIP L'AUTEUR CONSERVE LA PROPRIETE


OF TIIE COPYRIGHT IN HISfHER DU DROIT D'AUTEUR QUI PROTEGE
TIIESIS. NEITHER TIIE TIIESIS NOR SA THESE. NI LA TIIESE NI DES
SUBSTANTIAL EXTRACTS FROM IT EXTRAITS SUBSTANTIELS DE CELLE-
MAY BE PRINTED OR OTHERWISE CI NE DOIVENT ETRE IMPRIMES OU
REPRODUCED WITHOUT HISfHER AUTREMENT REPRODUITS SANS SON
PERMISSION. AUTORISATION.

ISBN 0-612-05815-8

Canada
• Dp<!icat,{:<! to 1lI}" fathcL Qingming \\"u, my mothcL Hanfcn
Hllang;, 1lI}" \Vifc, Sumillg Su, and my SaD, Di \\"u


• ~urfacc
ABSTRACT

The e1cctric Inining shovcl b the tll<lin rock loading ill<lcililll' al the ll1ajlH'ity ,If
n1ining opcratiotl~. Thc' performance..' a.nd lltilizaliotl ,)f thes,' tl1111ti-ll1illi'\1l
dollar Inachincs can go\'crn an entire tnin:ng. operat.ion. 1)('spitt.' tla'ir si~llilicall"t'.

they remain Jargcly unexplored iu ternIS of their potentia!. Th., thesis discusses th,
devclopmcnt of the forward and inverse kinemat.ic !Ilodds, t.he l\.'wton-Euit-r dynal'"
ics, and the actuator dynamics for an c!ectri,: mining shove!. The t.hesis n'Iat.,s 10.>

work undertaken t.owards the devclopment of a. mining shovd simulat.or, t.11t' purpoSl'
of which is 1.0 explore productivity and machine performancc issues ilS a funct.ion of
the interaction of the shove! with the groulld. Both the parametric and 1I0llpara-
mctric system identification resuIts arc presented, based on field t.est da.ta collect,'d
from an operating mining shove!. Simulator results describing analytical and ('!Ilpir-
ical models, as weil as continuous-time controller and disturbance reject.ion are also
presented. The thesis concludes with il discussion of planned continuing work,

• Il
• la
Ll {wllt'
AB5TRACT

I111"CéltlÎ(pH' ('·jc·('triqlH' repri·S('Il1.<' rOllt il

dans la plllp;lr1. dc's op(~rations dt' surfa('(' daIls \e c1oTllaillt' rninicr.

I){'rforlll<ttlc(~
de cliargeIllf't1t

de n's machines p01lvant. valoir jusqu·à. plusieurs rnillions de dollars


le plus utilis(',
L ·ut.ilisation ct.

(H1l. 1I11 impact. d<"~terrnillélllt sur les activités d'une rnine. i\lalgré leur rôle crucial! leur

pol.('ul.id n'sl.e largemeut iuexploité.

L'ohjet de cet I.e thèse cousiste il discuter le développement de modèles de cinématique


iuverse. de la dyuamique Newton-Euler et de la dynamique des actuateurs d'une pelle
él<'ctrique. La thèse relate le travail entrepris afin de développer un simulateur de pelle
nu;canique électrique dont le huI. est d'étudier la productivité et la performance en
fonction du degré d'interaction de la pelle avec le sol.

On présentera ki les résultats de l'identification paramétrique ct non-paramétrique


du syst.ème bas,x, sur l'analyse des résultats expérimentaux. On présentera également
les résultats provenant de simulateurs fondés sur deux approches différentes, l'une

• analyt.ique ct l'autre empirique. Les deux modèles utilisent des méthodes de contrôle
eu temps continu ainsi que des techniques de rejet des perturbations. La thèse se
t.ermine avec un aperçu des développements à venir.

• Hi
• ACKNOWLEDGE;\IENTS

1 wish 1,0 express IllY sinccrc gratitude t.o my thesis sllp(·rvisor. Pr\lfl':-;'SO!" 1.;\('t'q11t'

K. Daneshmend. for his continued support (from M.En!,!;. to Ph.D.). conlid<'nC<'. <lnd
guidance during the course of the investigation.

! \Vish to thank m" cosupervisor. Professor Car! Ilandricks. for his <ldvic<' and
help. Particu!ar!y, he taught me mining kno\\'!edge and ga\'l' me some "alnahle minin!,!;
materia! \Vhich had great help to me, a student \\'ith electrical enginl'Nin!,!; backgronnd.

1 \Vish to thank Professor Malcolm Scoble for his a.,sistance and advin'. 1h-
actually introduccd me to my first open pit mine. the asbcstos mine, in Black Lak<"
Quebec.


Financial support for this \Vork \Vas provided by a Strat.egic Projed Grant. from
the Natural Sciences and Engineering Research Council of Canada. 1 \Vish t.o thank
the above three professors for arranging my financial snpport from this gmnt..

1 \Vish to thank Professor N. Hori, for taking time to discuss sorne intcrcst.ing
issues about the research \Vith me.

1 wish to thank P&H Ltd. for providing much valuable data and mat.eria\.
could not have completed this work without this collaborative a..sistancc.

1 would also like to thank the staff and students at the Canadian Centre for An-
tomation and Robotics in Mining (CCARM), and at the McGill Research Centre for
Intelligent Machines (McRCIM), for providing excellent facilities and a stimulating
research environment. In particular, 1 wish to thank ProU. Mossor, Mr.C. Abou-
jaoude, Mr.M. Amjad, Mr.X. Huang, Mr.X. Luo, Mr.P. Knight, and Mr.A. Malnf, for
their various help. 1 also thank Mr.E. Garant, who translated my English Abstract.
to French Abstract.

• IV

Fi ,,;..!Iy. l ",isli 1(, t Ii""k IllY f"ti",r. Qing;lIli"g WIl. I"'Y mot !rer. Ilanfen
II1IrLllp;. ror I.lwir IOIl~ l.iIll(, (·duccd.ioll and encouragement. :\Ieann-hilc. 1 wish 1.0
t!r""k illY ",if". Snllli,,!; Sil. aI,,1 illY son. Di WIl. for t!r"ir 10\''', Ilnderstanding and
pal.i{~:IC(~.

• v
• CLAIM OF ORIGINALITY

The following original nmt.ribut iOIl:" Wl'n' tl1cHlt'.

• Systcnlatic investigation of the t-..lcctric tnining sho\'elllsil1~ fl\hot.ics and ('lml.rt)l

theory

• Creation of a model of the entire shO\'('\, including for\\'ard and ill\"'l's,' kinl"
maties..Jacobian, rigid body dynamics. and actuator dyuamics

• Complete analysis of the friction and impact b"t\\'(',·u huck<'l. and lIlul'kpi!<·. and
its effect on the shove! and its aduatol's

• Use of nonparametric system identification techniques to analys,' th,' h"havionr


of the shovel digging proccss

• Establishment of both analytieal and empirical shovel simulators. and validation

• of the overall system model using empirica! data.

• vi
• Contents

Chapter 1 1

Introduction 1

1.1 Overvic\\' . 1

Prcvious Shovcl Rcsearch . 2

I.:l Research Goals -1

1.'1 Thcsis Outline . 6

• Chapter 2

Forward and Inverse Kinematics of Shovel


7

2.1 Forward Kinematics of Shovel i

2.2 Inverse Kinematics of Shovei . . . . . . . . . . . . . . . . . . . . . .. 13

Chapter 3 17

Shovel Jacobian 17

3.1 Lincar and Rotationai Vclocity of Rigid Bodies . li

3 .) Jacobian . 20

• vii
• Chapter 4 .).)

Rigid Body Dynamics

4.1 Forward Reeu rsion for 1hl' V,'loci 1 jl'S and :\ eCl'll'ra 1ions "f 1 hl' ,\"i nls ~:;

4.2 Backward Rccursion for Ihe Ca!cnlalion ùf th,' I,'orel's ;\!Id 'I~lrqlll'S . :;0

Chapter 5

Analytical Actuator Dynamics and Shovel Parameters 33

.5.1 Analytical Actuator Dynamics . :1:;

.5.2 DC Motor Paramctcrs . . . .. :lli

Rigid Body Mcchanical Paramctcrs ·Ill

• Chapter 6
Rigid Body Incrtias . . . . . . . . . ·Ill

46

Friction and Impact between Bucket and Muckpile 46

6.1 Friction

6.2 Impact. ·1!J

6.3 Ail External Forces and Thcir Caleulations . 51

Chapter 7 54

Data Acquisition and System Identifications of Actuator Parameters 54

i.1 Overview of Data Acquisition " 54

• Vlll

-
1.-
'j Syslt'III Idf'I1t.ificatiollS of ï rflIlSf('r FtltlcliOllS

7.:1 Sysl.<'111 Id"lItil;"ilti""s "f Still.<' Spil,.!: ~I"dds G·I

7,·1 ParalJw1.rÎc COillpal'isOIl .

Chapter 8 75

Nonparametric System Identification of Digging Pro cess 75

S,I Elupiri"al Transfer-Fllnct.ion Estimation (ETFE) , 75

s pcct.,'al Allalysis Estimation (SAE) , i6

S,:I Lag Window Choicc i6

SA Noise Spect.rllm Comparisons of Armature Currents with and without


Digging , . . . S3

• 8.5

8.6
Noise Spedrum Comparisons of Armature Currents \Vith Easy and
Hard Digging .

Noise Spedrum Comparisons of Positions with and without Digging.


Si

Si

Chapter 9 92

Simulating Methodology and Simulator 92

9.1 Overal1 MachinefProcess Model 92

9.2 Continuous-Time Control1er 93

9.3 Saturation 104

9.4 Filter . . . 104

105


9.5 Disturbance

ix

9.0 .\llalytical :-lod,,1 Silllldalioll . Ill.',

9.ï Elnpirical ~Iodd Simulatioll of Ea:--y Di~~itl~ Ill'>

Elllpirical :-Iode! Sillllllarioll of liard I)i~~i,,~ . 1Ill;

9.9 Diggabiliry I"d,,)( . I~ll

Chapter 10

Conclusions and Suggestions for Future Work l ')"


"'
10.1 Conclusions . ,-,
.)-

10.2 Suggestions for Future \York 12S

Appendix A: Simulink Simulators 1:\0


Bibliography 1.';2

• x
• List of Figures

1.1 Typicill Hydraulic Mining Sho"c1 (Source: Demilg Corp.) 2

1.2 Typicill Electric Mining Sho"c1 (Source: P&H Inc.) 3

1.:3 Reseilrch GOills ilnd Steps o . . . . . . . . . . . .. 5

2.1 The Coordinilte Set Diagram. . . . . . . . . . . . . . . . . . . . . .. 8

5.1 Circuit Diagram for DC Shunt Motor . 33

DC Shunt Motor System Block 35

• .5.3

5..1
Rectangular Prism

Saddle Block . . .
41

42

6.1 Forces Acting on Shovel Machine . 4i

6.2 Impacts Acting on Shovel Machine 50

i.l Data Acquisition System for Shovel Machine . 55

System Identification Procedure of Transfer Functions . 5i

i.3 Pole-Zero Plot of the Swing Motor . 58

i.4 Frequency Response Plot of the Swing Motor in Discrete-Time 59

i.5 Frequency Response Plot of the Swi~ Motor in Continuous-Time 59


i.6 Pole-Zero Plot of Hoist Motol' . . . . . . . . . . . . . . . . . . . . 60

xi
• './

ï.S

i.9
l'req uenry Response l'lot of 1I0ist :\lolor in Di"T('k-Ti nl<' "

l'requenry Response l'lot of the lIoist

Po[c-Zero Plot of Cro,,"d i\lotor .,


~[otor in Continuons-Tilllc

.
. . li 1

li 1

(
..
)

'-
i.10 Frequency Response Plot of Cro,,"d Motor in Discrl'l.c-TinH" li:1

i.Il Frequency Response Plot of Cro,,"d Motor in Coutinuous-TiuH" li:1

i.12 Frequency Response Plot of the Swing l\lotor iu Disrret.(~Til1l(' Iii

i.13 Frequency Rcsponse Plot of the Swing Motor in Continuous-Tillle (ii

i.14 Comparison Between the Actual and Simulated Rotationa[ Anglcs of


Swing Motor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. ()~

i.15 Frequency Response Plot of Hoist Motor iu Discretc-Time iD

i.16 Frequency Response Plot of the Hoist Motor in Coutiuuolls-Time iD

• i.1 i Comparison Between the Adual and Simulated Rope Positions of Hoist
Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

i.18 Frequency Rcsponse plc·t of Crowd Motor in Discrct<..~ Time


il

i.19 Frequency Response Plot of Crowd Motor in Cont:.luous-Tirne

i.20 Comparison Between the Actual and Simulated Arm Extensious of


Crowd Motor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. i:l

8.1 Transfer Functions of Hoist Motor Armature (ETFE) . ïi

8.2 Transfer Functions of Crowd Motor Armature (ETFE) 77

8.3 Transfer Function of Hoist Motor Armature (SAE) 78

804 Noise Spectrum of Hoist Motor Armature Current (SAE) 78

• xii

,~"=:i '['r<m,fer Furlction of Crowd ;'vlotor ,\rmatllre (SAE) . . . . ...... i9

H.G Noi,e Spectrllm of Crowd Motor Armature Current (SAE) · .... i9

8.7 Noi,e Spcdrllm of Hoist Motor Armature Current (M=lOO) · .... 80

S.S Noise Spectrum of Hoist Motor Armature Current (;\'1=1.50) · .... 81

8.9 Noise Spectrum of Hoist Motor Armature Current (M=200) 81

8.10 Noise Spectrum of Crowd Motor Armature Current (M=100) . 82

8.11 Noise Spectrum of Crowd Motor Armature Current (M=150) . 82

8.12 Noise Spectrum of Crowd Motor Armature Current (M=200) . 83

8.1:J Noise Spectrum of Hoist Motor Armature Current with Digging 85

8.14 Noise Spectrum of Hoist Motor Armature Current without Digging 85

• 8.15 Noise Spectrum of Crowd Motor Armature Current with Digging

8.16 Noise Spectrum of Crowd Motor Armature Current without Digging .

8.1i Noise Spectrum of Hoist Motor Armature Current with Easy Digging
86

86

88

8.18 Noise Spectrum of Hoist Motor Armature Current with Hard Digging 88

8.19 Noise Spectrum of Crowd Motor Armature CUITent with Easy Digging 89

8.20 Noise Spectrum of Crowd Motor Armature CUITent with Hard Digging 89

8.21 Noise Spectrum of Hoist Motor Rope Position with Digging ·. 90

8.22 Noise Spectrum of Hoist Motor Rope Position without Digging . 90

8.23 Noise Spectrum of Crowd Motor Arm Extension with Digging 91

8.24 Noise Spectrum of Crowd Motor Arm Extension without Digging 91

• xiii
• 9.1

9.2

9.3
Simulation System St.ruct.ure

Actual and Simulated Rotat.ional Angles of Swinp;

Actual and Simulat.ed Volt.ages of Swing ~Iot.or


~lot.oc

...
.

lOi

lOi

9.4 Actual and Simulated Rope Positions of Hoist Motor

9.5 Actual and Simulated Voltages of Hoist Motor lOS

9.6 Actual and Simulated Currents of Hoist 1\'lotor 109

9. i Actual and Simulated Arm Extensions of Crowd Motor 109

9.S Actual and Simulated Voltages of Crowd Motor 110

9.9 Actua! and Simulated Currents of Crowd Motor lia

9.10 Actua! and Simulated Rotationa! Angles of Swing Motor (Easy Dig-
ging) III

• 9.11 Actua! and Simu!ated Voltages of Swing Motor (Easy Digging)

9.12 Actual and Simu!ated Rope Positions of Hoist Motor (Easy Digging)
111

\ 12

9.13 Actual and Simu!ated Voltages of Hoist Motor (Easy Digging) 112

9.14 Actua! and Simu!ated Currents of Hoist Motor (Easy Digging, Il:3

9.15 Actua! and Simulated Arm Extensions of Crowd Motor (Easy Digging) 113

9.16 Actua! and Simu!ated Voltages of Crowd Motor (Easy Digging) Il4

9.1i Actual and Simu!ated Currents of Crowd Motor (Easy Digging) ll4

9.18 Actua! and Simu!ated Rotational Ang!es of Swing Motor (Hard Dig-
ging) li!)

9.19 Actual and Simu!ated Voltages of Swing Motor (Hard Digging) 115

• xiv

!J.:W Actllal alld Simulal.ed Rope Posil.ions of lIoisl. :-'lot.or (Hard Digging) 1i6

!).:! 1 Acl.ual alld Simulal."d Voll.ages of lIoist Mol.or (lIard Digging) 116

!J.:!:! Actual alld Simulated Ciments of lioist Motor (Hard Digging) Iii

!J.2:1 Actual and Simulated Arm Extensions of Crowd Motor (Hard Digging) IIi

!J.2:1 Actual and Simulated Voltages of Crowd Motor (Hard Digging) liS

!l.25 Actual and Simulated Currents of Crowd Motor (Hard Digging) liS

!J.26 Main Variables of Electric Mining Shovcl . . . . . . . . . . . . . 119

9.2i Actual Current and Voltage of Hoist Motor in a Easy Digging Phase 124

9.2S Actual Current and Voltage of Hoist Motor in a Hard Digging Phase 124

9.29 Actual Current and Voltage of Crowd Motor in a Easy Digging Phase 125

• 9.:30 Actual Current and Voltage of Crowd Motor in a Hard Digging Phase 125

9.:31 Distribution of Diggability Indices

A.1 Electric Shovel Main System.


126

131

A.2 Td4 subroutine 132

A.a f4a subroutine . 133

A.4 v4ca subroutine 134

A.5 v4a subroutine 135

A.6 Tda subroutine 136

A.i na3 subroutine 13i

."l.S N33 subroutine .. .. . .. . .. . .. . .. . .. .. .. .. .. .. .. . .. .. .. .. .. .. .. .. .. .. .. 13S

• xv
• .-\.9 1'142 sllbrolltinc

:\.10 v4cl sllbrolltinc ..............................

:\.11 \'41 subrolltinc


1·lll

Il \

:\.12 Td1 subrolltinc 1·1 :!

A.13 n1:3 sllbrolltinc

A.14 P23 sllbrolltinc .

A.15 v4c2 sllbrolltinc

A.16 v42 sllbrolltinc Wi

A.1i n21 sllbrolltine I·li

A.18 n31 sllbrolltine 148

• A.19 n32 sllbrolltinc 1·1 !)

A.20 n22 sllbrolltine 150

A.21 f33 sllbrolltine . 151

• xvi
• List of Tables

2.1 I.i n k l',-<callleters 8

ï.1 l'ararnet.ric Cornparison of Swing, HoisL and Crowd Motors ..... ï4

!J.l SlImmary of Diggability Indices 1')')

9.2 Diggability Indices of Easy Digging 2 122

9.:l Diggability Indices of Easy Digging 3 122

9.-1 Diggability Indices of Hard Digging 2 . 1')')

• 9.5

9.6
Diggability Indices of Hard Digging 3

Diggability Indices of Hard Digging 4

9.ï Diggability Indices of Hard Digging .j


. .

..
.
. .
123

123

123

• xvii

Symbols

A coC'f[1cictlt l11atrix of stat.<..... spaCl~ equat.ion

ai dist.ance from Zi 1.0 Zi+1 meastlrt'd along .\i

B cocrRcicnt nlcl.trix of stat.e-space equat.iotl

BI coefficient matrix oi statc-spacc <'quation

b mechanical damping constant of DC motor

be c!ectrical damping constant of DC motor

b·• viscous damping coefficient

C coefficient matrix of state-spacc equation

friction coefficient of the crowd OC motor


Cl

C2 friction coefficient of the hoist OC motor

C3 friction coefficient of the swing OC motor

CI cos(Otl

C2 COS(02)

C3 COS(03)

D Maximum digging distance

Dl bucket length

di distance from Xi 1.0 X i+l measured along Zi

d4 translation of dipper handle

d4 translational velocity of dipper handle

• XV111
• F irnpact forces

FI i rn pacts along x:;

Fil impads along =5


F: llax rnaximum impact

1 Fi incrtial force acting at the center of mass of each link

f friction forces

fr payload gravity

f. bottom surface friction of outer bucket

• h
f.
bottom surface friction of inner bucket

two-side surface friction of outer bucket

two-side surface friction of inner bucket

if; force vector exerted on the ith joint

G( q) transfer function

GN(é') estimation function

GN(é') estimation function

9 gravity acceleration

H actual depth of material in bucket

Hf. final average depth of material in bucket


1 unit matrix

xix
• lm sum of the actllator and gearbox in('rtia~

la actuator incrt.ia

19 gearbox inertia

c'Ii inertia tensor

[xxi scalar clement of inertia tensor

lyyi scalar element of inertia tensor

l==i scalar element of inertia tensor

la armature current of DC motor

Il field current of DC motor

Il total current of DC motor

• J Jacobian

[( coefficient matrix of state-space equation

kp passive pressure coefficient

kr factor due to gravity( from nomograms)

1 1
[(1 M, + M,
[(2 deformation coefficient

kii gain(i=l ,2,3,4,5,etc.)

kl torque constant of swing motor

k2 back emf constant of swing motor

k3 torque constant of hoist motor


~ back emf constant of hoist motor

xx
• 1.:" torque con:-itant of crowd rnotor

l,., !J"ek ernf of crowd rnotor

1. lip friction force

IJfl arrnature winding inductance

L/ r.cld winding inductance

{.Ii length of link i

Nh hllcket mass

M,. payload

M, bllcket maso

M2 muckpile mass


nti rigid body i mass

iNi inertial torque acting at the center of mass of each link

n length number of bucket

ni number of teeth or lips

in-1 moment vector exerted on the ith joint

P payload

Pp lateral thrust per length of wall(Nfm)

Pr X coordinate of the cutting head frame {5} relative to frame {D}

Pu y coordinate of the cutting head frame {5} relative to frame {D}

P: Z coordinate of the cutting head frame {5} relative to frame {D}


ip.1+1 vector from the origin of frame i toward the origin of frame i +1

xxi

ip.
Cl center of ma~~ for each lillk

Q controllability matrix

Qi gcon1ctry constant

q shift operator

R resistive force on blade

Ra armature winding resistance

Rf field winding rcsistancc

l+1 R rotation part of the coordinate i + 1 with re~pcct 1.0 tl\(' l'nordi liaI... i
ri gearbox ratio

S Laplace transform

• SI sin( 01 )

T tooth friction force

Tm torque generated by the DC motor

Tdi disturbance torque 1.0 the DC motor

:-lT transformation from frame {il 1.0 the frame {i -I} for cach link

u( t) system input

li,. bucket motion velocity


muckpile velocity

xxii
• , iupul Voll,,!;,· of D(, ulOtor
"
ri!;id IJOdy i+ l 's lilll'"r \'('Iocity \\'ilh respecl 10 rigid body i

..
"H-I ri!;id body i+ l's liuear "ccc1eraliou \\'ilh respeci 10 rigid body

Iillear "cce1eralioll of Ihe Cellier of lIlass of each lillk

\\'eighlillg fllnction

w blade width

width of the lips

\\'idth of the tccth

.\"" X coordinate of frame 6 origin relative 1.0 frame 0

• X2

X3
state variable representation of 0 for swing motor

state variable representation of w for swing motor

state variable representation of i a for swing motor

X.I state variable representation of 0 for hoist motor

;r a state variable representation of w for hoist motor

Xt; state variable representation of i a for hoist motor

X7 state variable representation of 0 for crowd motor

Xs state variable representation of w for crowd motor

X9 state variable representation of i a for crowd motor

xi steady-state values of state-space variable

• xxiii

~.rn

l:\"

}o y coordinat.e of frallle (i orif!;În !"l'Iat.i\"<, t.o fralll<' II

Y( 1) syst.em ou t.put.

t:>Yn

Zo Z coordinate of frallle GorigÎn rdal.in' 1.0 frall1<' II

i+IZ"'
:.Jr+l along the Z-a.,is direction

destination of buckct digging

(XWY60':W) start of bucket digging


Greek letters

Qi-1 angle betwccn Zi and Zi+1 measured about Xi

'Y unit wcight of the fiB material(Nfm)

<5i material dependent constant

o rotor position(radians)

Oi angle betwccn Xi_1 and Xi measured about hi

01 rotatory angle of swing motor

03 rotatory angle of hoist motor

Oi angular velocity of DC motor

Oi angular acceleration of DC motor


Ôi estimator

xxiv
• Il

tiu\(' ("OTl ..... taIlt.

l.iul(' COllstant.
hIl,,

l:;
,il"
, " (<-:) Iloise spcd.rulIl

'P digging angle

\ hlade t.ip dl'pt.h

<-: "-1.'/'\'
_'1 . • • 1..
. -_ 1 ~ ...•.1\'

V.Ji angnlar vcloeit.y of OC mot.or

l:.l.,,'lr referencc angular vcloeit.y for swing motor

u,,'2r rcfercncc angular vclocity for hoist motor

• W3r

' Wi + t

i•
Wi+1
rcfercnec angular vclocity for crowd motor

rigid body i+l 's rotationa! vclocity wi~h rcspect 1.0 rigid body i

rigid body i+ l 's rotationa! accclcration with respcct 1.0 rigid body i

• XXV
• Chapter 1

Introduction

This thesis is an application of robotics and control th..ory 1.0 th.. op..ration or t 1...
clectric mining shovel. lt is also a dctailed description 1.0 th,' ..1<-ctric Illinin!,: sho\'l'I
from the points of view of mining, electricity. and Illechanics.

1.1 Overview

Surface mining represents one of the world's largest industries. The Illining shov..1

• plays a key l'ole in many surface mining operations. It is used as a prodnct.ion ma-
chine in conjunction with drill and blast operations to Illove waste rock or ore. Th"
rate and efficiency with which mining shovcls operate in loading haul trucks will
largcly determine the efficiency of the overall mining operation. lIence, most surfac,'
mining operations tend to evaluate their productivity based on the perforuHLllce of
their primary loading equipment. There exists considerable scope for prodndivity
enhancement in surface mine extraction through the optimized lise amI operation of
current excavation equipment.

There arc two types of mining shovcls used in current mining indnstry, i.e., the
hydraulic shovel and the e1ectric shovel.

The typical hydraulic shovel is illustrated in Figure 1.1. The machine has versa-
tile featurcs, and it is welI adapted to selective mining mcthods. It has a wrist action
of the dil'l'cr, and is able to separate waste from ore in order to reduce the l'ost of
ore processing. Although the hydraulic shovel has many merits, the electric shovel is

• preferred in the hard digging environments, due to its power and capacity.

1

l "OWC........ T
:'IJfL l'ANK
J OlIllSElltvC)lA
.. CONtPOl, Yal,v(S

Figure 1.1: Typical Hydraulic Mining Shove1 (Source: Demag Corp.)


An typical c1ectric shovc1 is illustrated in Figure 1.2. The machine consists of
two major sub-systems: the mechanical and electrical systems. From a mechanical
viewpoint, the shovc1 has thrcc main parts: the crawler and gantry structure; the
boom and rope assembly; the di l'l'cr handle and dipper. Corresponding to the three
ll1echanical parts, there are also thrcc major e1ectrical parts: the swing motor, which
drives the gantry structure rotation; the hoist motor, which drives the rope assembly
payout and retraction; and the crowd motor, which drives the linear displacement of
the dipper hanclle and dipper. This thesis will focus on the e1ectric shoveI.

1.2 Previous Shovel Research

Rcsearch of the actual performance of an e1ectric mining shove1 was initially under-
taken by Keller (19;8). His system used a microprocessor based monitoring system to
enhance the performance of c1ectric mining shovels in the coal industry. Keller's work
included shovel monitoring, but did not relate to data analysis and shovel responses

• 2

Figure 1.2: Typical Electric /Vlining Shovcl (Sonrce: l'kIl Ine.)


to variations in muckpile diggability. Williamson ct al.( 1!J8:l) and Mol et al. ( 1!J8ï)
undertook investigations into muckpile diggability by monitoring shovcl performance.
Their work measured blast efliciency through the performance monitoring of iLll clec-
tric mining shovel. Their work was undertaken in separate uncont.rolled gcologic
and blast environments, and did not account for several llncontrollable variahles.
In addition, no data on dipper trajectory was recorded assist in assessing mnckpilc
diggability. Carl Hendricks et al. (1989) investigated another clectric mining shovd
operation. This work included geologic investigation and blast design, post hl,,-,t eval-
uation, data acquisition, and analysis of monitored shovcl performance paramcters.
This research demonstrated the rclationships between shovcl performance and lll11ck-
pile diggability and the significance of the digging trajectory. Althollgh there hiL~ not
been any research directly re!ating to the friction of shovcl buckets so far, Rowlands'
thesis (1991) investigated the friction of a dragline bucket. His work provided some
basis to our research on the friction mode! of a shovcl bucket. Hadjigeorgiou .1. ct al.
(1993) deve!oped an empirical classification system which was used to asscss the ea~e


of excavation. He took inertial effects into consideration in his model.

3

1.3 Research Goals

No l'ri or \\'ork h'L' 'L"ernbled a total system rnodel of an e1ectric shovei. Our \\'ork
foe\lses on rnathematic TT10dclling of the system. It uses up-to-date robotic and auto-
matic techni'Jl"'s 1.0 solve sorne very practical issues, su ch as the rclationship bet\\'een
digging and the De motors of the shove!. ;\'Ioreover, advanced system identification
and spectrnrn analysis techniques are used 1.0 analyze data collected and 1.0 assess
rock fragmentation and diggability. The overall goal of the research is 1.0 increase the
prodnctivity and raise the efficiency of clectric shovcls through a better understand-
ing of the machine operational limitations and its interaction with the rock. The
methodology that is proposed rclies upon both analytical and experimental studies of
the behaviour of the machine and its performance while digging. In the scope of this
thesis. these studies arc rclated 1.0 the model of the complete shovel, including for-
ward and inverse kinematics; Jacobian; rigid body dynamics; actuator/transmission
dynamks; full analysis of the interactive effect of friction and impact between bucket


and muckpile; introduction of the parametric and nonparametric system identifica-
tion to the shovcl and its digging process; and establishment of both analytical and
empirical modcl simulators.

The research is aimed 1.0 advance mining technology in two ways. In the medium
term, the thesis should assist the optimization of shove! productivity. In the long
term, however, it should contribute 1.0 the fully automated operation of the electric
mining shovcl with a computer vision guidance system. Overall system model could
contribute 1.0 computer control design which would make a shovel fully automated.
Kinematics modcl could contribute 1.0 trajectory planning/task planning which would
enable the shovcl 1.0 follow a human operator's guidance. Figure 1.3 summarizes the
research goals and steps.

The thcsis investigation includes three phases: analytical modelling; empirical


rnodelling; and system simulation. In the analytical phase, mathematical rnodeJs
pararneterized with actual data from a P&H 2300XPA shovel were huilt (Help was


receivcd from the P&H Company, see Chapter 5). In the empirical modelling phase,

4
• 1 Elcctric Mining Shovel Rcscarch

1 McdiumTcrm 1
~>
1

, , t
Image Processing Field tcsting
MYTHESIS f-
of fragmentation
Ph.D. thesis
1 1
Othcr tcam
rncmbcrs
1 and P&H data

of Mr. A. Bedair
:...........................................................u············ ........................................................................... ....

'-- Analytical
t
Empirical

Modelling

t
1
, Modelling
1

1 Machine
Muckpile

t +
1
, , 1
+
Multibody Actuators Friction &


Kinematics Jacobian Impact
Dynamics Dynamics
1
t
Overall MachinelProcess Model l-

SCOPE OF THESIS
I ,
Simulation
1
1

Performance
Analysis
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .u ••••••••••••••••• u •••••••

~
LongTerm
t
1
, t
1
Trajectory Planning 1 1
1 Computer Vision 1Computer Control
Task Planninl!

+
Fully Automated
ShoveI

• Figure 1.3: Research Goals and Steps


5

field \,('sts and data acquisition were undertaken. The parametric and nonparametric
system identification W'cs implemented so that an empirical modcl could be IHlilt. In
t.he simulat.ion phase. both the ,malyt,ical and empirical modcls were set up and used
t.o si,nulate and validat.e the c1ectric shovcl operation.

1.4 Thesis Outline

In Chapter 2. the forward and inverse kinematics which establish the rclationship
betwccn the bucket and ail joints arc obtained. This rclationship lays a foundation for
t.he bucket tracking calculation and trajectory contro!. The shovd Jacobian, which is
t.he foundation for shovcl specd calculation and control, is included in Chapter 3. In
Chapter ·1, the rigid body dynamics arc calculated. The dynamics enables us to re!ate
the digging forces and rigid body motion forces to the load torques of the shove! actu-
ators. Chapter.'j examines the shove! actuators and their system parameters, which


are components of our simulators. Friction and impact are investigated in Chapter 6.
Based on Rowlands's work, we deve!op a more complete friction model of the shove!
bucket. In addition, the impact effects of the bucket encountering large fragments
are investigated. The experimental investigation and parameter system identification
arc described in Chapter Î. Bath the least square identification for actuator transfer
function and the ma.ximum like!ihood system identification for actuator state-space
function are presented. Chapter 8 gives the nonparametric system identification to
the analysis of the armature currents with and without digging, as weil as easy and
hard digging. Morcover, the position noise with and without digging is aIso depicted.
Both the analytical and the empiricaI model simulators are developed in Chapter 9.
Final simulation results are aIso presented in this chapter. Conclusions are given in
Chapter 10, which also reviews achievements and suggestions for future work.

• 6
• Chapter 2

Forward and Inverse Kinematics of Shovel

Kinematics is conccrned with defining an object.'s mot.ion. i.". an o"ject.'s posit.ioll.


orientation, vclocity and acceleration. in t.erms of geomctrical propert.ies. Sinn' t.1",
joints in the shovcl arc connected 1.0 each other, t.he mot.ion of one joint. will alf,'ct.
the motion of the other joints. Kinematics consists of forward kinelllatics and inverse
kinematics.

2.1 Forward Kinematics of Shovel

• The forward kinematics of a shovel nccd 1.0 be derived in order 1.0 solve the t.rajedory
generation problem of the machine. Its trajectory is dependent on the swing, hoist. and
crowd motor motions. The trajectory also contributcs 1.0 the calculation of fridion
betwccn the bucket and muckpile. This issue is considered in detail in Charter 6.

The forward kinematics problem is as follows: given the joint angle vect.or [Oh 0"
(}3, d4 ], find the position and orientation of the end effector of the shovcl in the work
location coordinate system, which has bccn determined with respect 1.0 a referencc
coordinate system. Figure 2.1 is the coordinate set diagram of the shovel with severai
coordinates fixed on it. If one considers the base of the machine to be lixed when the
shovel is digging, the frame {O} is the world coordinate frame; frames il}, {2}, {:l},
{4}, and {5} arc fixed on the shovel and the frame {6} is lixed on the location where
the shovel is working.

In terms of the Denavit-Hartenberg notation, the shovcl can be described kin<.~

• matica11y by assigning four values 1.0 each link. Two values arc for the link and the

ï

T 'n
1
1 - - - - - - - '", -------1


Figure 2.1: The Coordinate Set Diagram

1 O'i-1 ai-l di 0;

1 0 0 0 01

2 90· LI 0 45·

3 0 L2 0 03

4 90· L3 0 0

5 0 0 d4 0

Table 2.1: Link Parameters

• s

other two \"alues arc for the cOllllt'ctioll bt'l\Iw'lI the lillb. l"sllally. 1111·e,· ,ariables.
Oi-l' ai_1 and di. arc fixcd link paranwt('rs but Oi is a yariablt' fol' rot.at.iOllal ca.:-'t'.

Thrcc variables. 0i_l' ai_l. and Di are fix(·d link parattleh'rs hut. <Ii i~ a \'ariahh· for
translational case. Ba...,cd on this foundat.ion. ont.' l'an detl-nnit1l' t.ht' slw\'('l's cOt,ll'di-
nate structure. Table 2.1 describes the lillk paramctt'rs. whert'

L, is the length of link 1;


L 2 is the length of link 2:
L 3 is the length of link 3:
L" is the length of link 4;
d., is the arm extension of dipper hanclle:
0, is the joint 1 (swing motor) rotational angle:
O2 is the joint 2 angle; it is fixed (0 2 = 45°).
03 is the joint 3 (hoist motor) rotational angle. corrcspollding 1.0 rape positioll.


Ob 03 , and d., arc variables; while other parameters arc all cOllsl,ants. The physk;'[
meanings of the parameters or variables arc also shown in Figllre 2.1.

Computing the individual transformations for each link which ddilles frame {i}
relative to the frame {i -I}, one l'an use the formula [12]:

( COi -SOi 0 fli_1

i-1T _
i - l";=H
SOiSOi_,
COiCOi_,

COjSOi_l
-SOï_1

CQi_t
-SOi_ld j

CO'i_1 (li
(2.1 }

0 0 0 1
thus,
CI -SI 0 0
s, Cl 0 0
~T= (2.2)
0 0 1 0
0 0 0

• 9
• = ·Vi o
Il''~illlS<' link 2 is fiX(·d. O2
LI _LI 0 L,
2 2

0 0 -1 0
.'.'1'= LI LI
(2.:3)
2 2 0 0
0 0 0

Ca -83 0 L~

$3 C3 0 0
3
2'1' -- (2.4)
0 0 1 0
0 0 0

1 0 0

3T _ 0 0 -1 o
., - (2.5)
o 0 a
o 0 0 1
1 0 0 a
a


JT = 0 1 0 (2.6)
o 0 1 d.,
a a a 1
MlIltiplying eqllation (2.2) by (2.:3), and we get
CI a-SI a Y,f _il a LI
2

?T ~T = SI c, a a a a -1 a
(2.i)
a a 1 a il
2
il
2
1 a
a a a 1 a a a 1
LI
Z-CI -Y,fCI LICI

il
.;" s} -ifs} LIS I
= il (2.8)
2
il
2 a
1

C3 a L2
-~~3

il.
2 St
_Jtl,
2 "1 -CI L,sI S3 C3 a a
?T ~T âT = (2.9)
il
2
Jtl
2
a a a a 1 a
a a a a a a 1


1

la

r; r;
,fl
~ CI('3 - :::L..i.c\
:.! •..;·\
• -Tcl . . ;t - /':>
Telc;1 - ..... , ~,
'.! (1 1. ~ + 1. 1"1
,fl ,fl '"
=
'2,- ..... I C3 - -:,- ""1·...;\ '" . . ;\ -
-7·.;1 ~·.;l(·:t -("1 T"1'~'';I + /. 1·... 1 \~,lll)
r:;
,fl
'2 C3 + ,fl
"1,- 8Jo -~ ... \
'2 '. + ~('.\
'2 • II 0:11.
'..! '.

0 0 II
Finally. multiplying (2.10) by (2.6). alld \\'t' gt'l
U'[' _ u'[' \'[' ~'[' "'['
·1 - 1· '2 a ·1

- ,2-C1""3
f l ,- f 2-C1('3
l -"1 ,fl 1
'2-C1''.! + 141("1 II o
- ,'2-8183
f l ,- f'2-·...
l 1C3 -CI '2 .'2,·...' 1 + l ~1·...' t
,fll n 0 -1 0
-f~3 + f C3 0 fl.~ II II II
o o 0 1 0001
-~1 . Z 1-'3 + 0:
,flc' e
'2- c t:~ ,fl,.
'..!·I 1-'2, + 1-1<~ 1 + 1-:l (,fl(. ('
'2 1:1 - ~("')
- '2
1·...:\

-Ct
,fl
:t . . . tS3+ ,:..t SIC:t
f l , f'2,-L'2,'''1
l + LI";;', + 1.:,( ,'2-f.... ,(';\
l ,- f l :,!- .... I .... :tl
o f~3 - fe" fl,~ + Id f"" + 'f"")
o o o 1
(2.11 )


Because

(2.12)

one has
1 0 0 0
o 1 0 0
~T - ~T
o 0 1 d.,
o 0 0 1
~CIC3 - ~CI~3 -SI il
2" CI 83 + il '2"C1C3 151

,fl. C
'2 .. 1 3 -
,fl•.
'2 "1 8 3 -CI il ... + il
'2 -"'1·...3
.. C
'2 ,"1 .:~ l52
(2.1:\)
~C3+ ~~3 0 il. ,Ji
'2'''3- ZC3 lr.3
o 0 0
where

(2.H)

(2.1'i)


(2.16)

11
• t.11f"f(·feJ("('
"'" ';1'
1; ;. -- ',l,'"

:. - "1'-1 "1'
"1'
(2,17)

(2.1 S)
" "
V,",.
a a .:ra
uT-
a a Yo
f' - (2.19)
a -1 a =0
a a a 1
",here {xu. Yu. =,,} i~ the origin of the coordinate {6} ",ith respect 1.0 the coordinate
{a}. x" =!J. y" = o. =u = -1.5,
a a Xo
1
-1
1 a a -xo

°T- I- a a Yo a a -1 =0
o - = (2.20)
a -1 a =0 a 1 a -Yo
a 0 0 1 0 0 0 1


Bence, one has

°T
.5
-- °T-IOT
6 5

1 0 0 -xo ~CIC3 - ~CIS3 -SI il


2 Ct S.3 +il
2 CIC3 t SI
a 0 -1 =0 il
~t$ICJ - il
2"$1$3 -Cl ~StS3 + 4SIC3 t S2
=
o 1 0 -Yo ~C3+ ~S3 0 4 S3 - 4C3 t53
o 0 0 1 o 0 o 1
~CI(C3 - S3) -SI 4CI(C3 + S3) tOI

=
-4(C3 +S3) 0 4(C3 - S3) t0 2
(2.21 )
~SI(C3 - S3) -Cl 4S1 (C3 + S3) t 63
o 0 o 1
",here,

/01 =
Vi Vi
2 cl(d4(S3+C3) + L3(C3- S3))+CI(L 2 2 +Ltl-xo (2.22)

t';2 =
Vi Vi
- 2- (d4(S3 - C3) + L 3(S3 + C3)) -
2- L 2 + =0 (2.23)
Vi Vi

• t63 - 2-SI(d4(S3+C3) + L 3(C3- S3))+St( 2-L2+L1)-yo

12
(2.24)
• Aft.er the abo\"C' leugthy algebraic l11aniplllat.ion. mil' C'lIl nht.ain t h(' huckt'l Ih)~i­

tion with respect to tl1l' work-site (lll11ck-pile) rderence coordinat,· {li).


knows t.he joint position 01.03.d.l. This is t.he forward killelllat.ics shown

x" = J2 v0
:2 CI (d., (S3 + C3) + 1.,,( c" - .,,,)) + c, (T I.~ + 1," ) -
il'

ilS

.1'"
ion)'; as
follo",s:
'li'"

( ')_._'
""1)

J2 v0
y" = -T(d., (S3 - C3) + L"(c,, + .,,,)) - TI.~ + =" _
( ') ....
"1')
)

=6 =
J2 s ,(d,, (S3 + COl) + L3(C3 -
T .,,,))
J2 + LI) -
+ SIr Tl,~ y" .,-)
( ')_._1

If 0, = 0, c, = L s, = 0, i.e. shovel st.ands on it.s original cent.ral posit.ion. We ha\'e

X6 =
J2
:2 (d., (s3 + C3) + L3 (c3 - SOl))
J2
+ Tl,~ + L, - .
.\"

Y6 -
J2
- :2-(d.,(S3-C3)+L3(C3+S3))- :2-L~+Zo
J2 (2.29)

=6 = 0 (2.:10)

2.2 Inverse Kinematics of Shovel

• The inverse kinemat.ics of shovel has not becn llsed in our syst.em simulat.ion. But.
because il. is a fundamental work t.o our fully-aut.omat.ed shovel in fu t. ure, wc st.ill
show its manipulation here.

The inverse kinematics problem of the shovel is as follows: given t.he desired
Cartesian position and orientation of the last link of the machine relat.ive t.o t.he
work location, calculate a set of joint angles which can achieve the desired posit.ion
and orientation. In order 1.0 solve this problem, one must first perform coordinat.e
transformation 1.0 obtain the bucket coordinate {5}, relative t.o the coordinat.e {l},
and then use the inverse kinematics 1.0 solve the joint angles.

Multiplying (2.3) by (2.4) and we gel.


lfi _lfi 0 LI C3 -83 0 L2
2 2

0 0 -1 0 83 C3 0 0
~T ~T= lfi
lfi 0 0 0 0 1 0
2 2

• 0 0 0 1

13
0 0 0 1
• 4lca - "3) -4('<3 - c:.) 0 lil
2 "-2 -'-
' 1...-)

0 0 -1 0
= (2.31 )
4{C3 + S3) 4{C3 - Sa) 0 liL
2 2

0 0 0
Multiplying (2.:31) by (2.5), and wc gct
q{C3 - S3) -q{S3 - C3) o qL 2 + LI 1 0 0 L3
Ir 327' 3'7'-
o 0 -1 0 0 0 -1 o
:.! ·1· - _~ _r.;
~(C3 + S3) ~(C3 - S3) o qL 2 0 1 0 o
o 0 o 1 0 0 0 1
qh - S3) 0 -q{C3 - S3) 4(C3 - s3)L3 + *fL 2 + LI
o -1 0 o
(2.32)
- q{C3 + S3) 0 -q{C3 - S3) *f(C3 + s3)L 3 + *fL2
o 0 0 1
Multiplying (2.32) by (2.6) and we get

IT 2IT 32T 3T 'IT


5 -1 5

1(C3 - S3) 0 -4(C3 - S3) t02 1 0 0 0


0 -1 0 0 0 1 0 0
-
*f(C3 + S3) 0 -4(C3 - S3) t 03 0 0 1 d4
0 0 0 1 0 0 0 1

*f(C:l- S3) 0 -~(C3 - S3) t l2


0 -1 0 0
- *f(C3 + S3) 1 -4(C3 - S3) t 13
(2.33)

0 0 0 1
Where,

t02 - (2.34)

t03 - (2.35)

t 12 - (2.36)

• -
t 13 (2.37)

14

Bccausc
-1
CI -$1 0 0 CI •... 1 0 II

01'-1_ $1 CI 0 0 - ·... 1 CI 0 0
1 - = ( _.'-'Is)
0) ..
0 0 1 0 0 0 0
0 0 0 0 0 0
01'-101'
0';1' = 1 5 (:!0:1!1)

Assuming
rll rt~ 1'13 Px
T:'H T22 T23 Pv
01'-
5 - (2.'10)
r31 T32 r33 P:
0 0 1 0
Where P"" Py , and P: are end effector positions with respect to the world coordinate
{O}. From (2038), (2.39), and (2.40), one has

CI SI 0 0 rll ri:! r'3 P'"


-SI CI 0 0 r21 rZ2 r23 Pv
511' -- (2.'11)
0 0 1 0 P:


r31 r32 r33

0 0 0 1 0 0 0 1
Equating the (2,4) element of equations (2.33) and (2.41), one has:

Py
O. = arctan- (2.42)
P'"
Equating the (1,4) element of equations (2.33) and (2.41), one has:

Ct P",
-12
+ 5t Py = -Td4(C3 -
-12
53) + 2 L3(C3 -
-12
53) + 2 L2 + L, (2.43)

Equating the (3,4) element of equations (2.33) and (2.41), one has:

(2.44 )

Subtracting (2.43) from (2.44), one gets

(2.4.5)

• 15

:\rritfl~ifl~ (:!.·1'j). 01'" l''L''

(2..16)

(2.·1i)

Traflsfcriflg (2.'16) iflto angle form. one has

(2.48)

Arranging (2.44), one gets

P: - 4(C3 + s3)L 3 - 4L 2
d., =
-4(C3 - S3)
(C3 + s3)L 3 + L2 - .Jip:
= (2.49)

In short, one can obtain the joint position Oh 03 , d'l as long as one knows bucket

• position P;"Pu,P:. This is the inverse kinematics shown as follows:

01

03
=
=
arctan Px

arctan
Pu

+ P: - CI Px - slPu
LI
.Ji2L
(2.50)

(2.51 )
3
(C3 + 83)L3 + L2 - .Jip:
d'I = C3 - 83
(2.52)

• 16
• Chapter 3

Shovel Jacobian

In order to analyze the motion of a manipulator. the ~.Jacohiall·· cOlln·pt. i, ("ommollly


utilized. The Jacobian de,cribcs the relation between the linear alld allgular velocity
of rigid body. It is also the basi, for the formulation of the ,hov"", dy"amic, whi("h
will be described in the next chapter. Furthermore. whell olle ("a!culate, the impa("t
in Chapter 6, one will use the shovel .Iacobian.

3.1 Linear and Rotational Velocity of Rigid Bodies

• The angular velocity and linear velocity propagatioll frolll joillt 1.0 joillt cali h,> ex-
pressed by the formulae [12]:

For a rotational motion,

.+1 z·
i+l
Wi+l = ii+lRiWi + O·
i+l i+l (:1.1 )

i+l vi+l =ii+l R(i Vi +i Wj x i p


i+l)

For a translational motion,

i+l i + i+IZ· i+l


Wi+l = i+lR
i Wï

i+l

i+l. _i+l R(i.


vl+l -i V, +i Wj x ip.)
1+1 + d'..Z·.1

From (2.2), (2.3), (2.4), (2.5), and (2.6), one can obtain
If -If
~R = 0 0 (3.5)
( il
• o
17
2
il
2

';'11= (~ ~ ~)
°°
llsing (:1.1) and (:3.2) for rotational motion,
1

'w,~ rD (a.S)

'V,~ m (a.9)

• (a.lo)

,/f °° (0)° + (0)


2V2 =
( -,/f
,/f)
,/f ( ° ) (0)
? (LI) °.I x =
(a.11)

° -1 ° ° ° -LIO
01

(a.12)

(3.13)

• 18
• (:1. 11 )

Using (:3.:3) and (:3.4) for translational motion


(:1.1(;)

• 19

3.2 Jacobian

TI", Jacobi ail is a Illllitidilllellsiolla\ forlll of 1. hc dcrivat.ive. In shove! ca$c. il. is a t.hrcc
dilllellsiollai deri\'at.iV('. Frolll (2.10). ulle kllo,,"s
-SI

(:U8)

Becallsc

(:3.19)

from (:l.Ii) and (:l.lS). one has


(3.20)

where.
J2 . J2 .
~';I = 2-C,((C3-S3)d.,+(C3+S3)L3)03+ 2-C1(C3+'<3)d'1

J2 J2 .
+ s,CCLi + 2 L2 ) - 2 L3(S3 - C3)01 (3.21 )

J2 J2
l'52 = 2- s,((d.,(C3 - '<3) + L3(C3 + s3);53 + 2- SI(C3 + s3)d.,

J2
+ c,((L, + 2- L2 ))
J2
2- L (S3 -
J2
C3) + 2- d4(C3 + S3))01
. (3.22)
- 3

V5J = J2
2 ((C3 + s3)d., + (S3 - c3)L3)03 -
. J2 .
2 (C3 - s3)d. (3.23)

Arranging (3.20). (3.21), (3.22), and (3.23), one has

°Vs =
('" X21
Xl:!

X22
' ') n
X23 03 (3.24)

• X31 X32

20
X33 d.

\\'hcrc.

h- ) r.;
v-
;rl1 = .<,(( 1., + -l."
2'
- -1."(-,,,
2"
- c'.))
.
(".)._,}
'I~)

h + (e" + .<:.)/.,,)
;r12 = TC1((C" - .<")11,, (:\.2li)

h .. .,-)
_,
~r13 = T'<' (C3 + .<,.) ('1

h h
.<,((L, + -l.,,) - -l.,.(-". - .."S)
.T21 = 2' 2"
c'.))
.
('1 _~

h + l."(c,, +-,,,))
X22 = T-,,((II'I(C3 - -,,,) (:1.2!1)

h
X23 = TCI(C3 + S3) (:\.:10 )

X31 = 0 (:1.:11)
h
X32 = T((C3 + S3)e/. + (S3 - 1 c,.)L,,) (:\.:12)

h
X33 = -T(C3 - S3) (:\.:1:1 )


The variables are the components of the Jacobian. The shovd .Jacohiall is

°J(O) = (::: ::: :::)


X31 X32 X33

Finally, one can get the rclationship betwecll the shovel's \inear vclocity and angu\ar
velocity
o 0 •
V:; = J(O)O

• 21
• Chapter 4

Rigid Body Dynamics

J)yn;llllics is the study of the rclationships bctween forces, torques, and motions. The
motion of the shovcl arises from torqucs generated by the actuators. and byexterna!
forces. The Newton-Euler formulation can be applied to calculate the torque vector.
T. if one knows the trajectory point. fJ, è. and ë. One can also calculate fJ, è, and
fJ, if one knows the torque vector, T. The vectors which locate the center of mass for
each link of shovcl are:

• = -L2
2p 2 •
1 Pel = OXI. c'"
• x•' (4.1 )

= L2 X3, 'Ip d.l •


3p 3•
c3 Col = 2 =-, (4.2)

The inertia is a very complex issue. Considering that each link has symmetry
(although there are sorne asymmetrical clements, compared to the major element, in
engineering, one can neglect the asymmetrical elements), one can choose the principal
,,-'(cs of inertia; therefore, the inertia tensors are as follows:

(4.3)

0 0
c, /3 =
C-'
0 /VfJ3
oo ) , <4/4 =
CT
/w4
1:) (4.4)

• 0 0 /::3

22
0
• The rotat.ions betw{·,'nlll'ighbouriug link frallles ar,' d,'scril",d hy f,)rIllllla"l:;.rl) .
(3.6), and (3.7).

By applying of the aua!ysis t.o t.he bllrkl't. (Chapt.l'r


one can considcr the cffccts of t11C' c:'.f.crnal forccs and
(i will intrnduCl' t.his issn,'),
tllOllll'Ilt.S act.itl~ on t.h... hl1Ckl'l,
as follows:

f~ = (~:) = (f~I), n~ = (::::) = (~)


f53 f53 "5:1 0

The complete algorithm for maniplilating shovd rigid body dyualllÏl's ,'onsists
of two parts: first, forward recursion for the vdocities and accderat.ious of the joint.s:
second, backward recllrsion for the calclllation of t.he generali~ed forces. They will \'"
discussed individually bdow.

• 4.1 Forward Recursion for the Velocities and Accelerations


of the Joints

For the shovel with four links, one l'an compute link i: 0-+4 [12]:

i+1w:
1+1 -
_ i+t Riw: +
i 1
ir1+1 i+1 Z·•1+1 (4.6)

i+1' i+1Ri ' +i+IRi


Wi+l = i Wï i
O· i+1Z' i+l
Wi X i+l
+0"i+1 i+1Z' i+l (4.7)

i+1 Vi+l
• = ii+l R(i Wi• X
i P i+l +i Wi X (i Wi X i p.1+1 ) +i v.
'.) (4.8)


i+1 P P c· 1 ) +i+1 Vi+1
i+1'
Vc•1+ 1 =i+1 Wj+1

X c· 1+
1
+i+1
Wj+1 X
(i+1
Wj+1 X
i+1
1+
• (4.9)

23
• i~lF
. j ..... l = 1ni+1
i+l·
· l
V c 1+ CUO)

(4.11 )

Let us cOllcret.ely manipulate t.he forward recursion of the shove!. The gravity
crfcct is calculated as:
O, •
vo = gzO (4.12)

The base of the shovcl machine is not rotating, therefore one has:

",'0 = 0, ';;0 = 0 (4.13)

The forward recursions for Iink 1 are as fol1ows:

(4.14)

• (4.15)

(4.16)

(4.17)

(4.18)

• 24

(.\.l!})

The forward recursions for link 2 are as follows:

2- _(_~
W2- 2
00 ~) (00) _ (~::I)
2 - 2 1 (·1.21)

• 'v, - (~ 0
°

4) (")
-1 ° 01 °

n (") Cl C')
+ ° ) -"-4 ° 4
= L/il = 4(L O<.+
I 9) (·1.22)
9 ° ° -1 9 -LIO I

(4.23)


The forward recursions for link :3 arc as follows:

• (4.26)

(1.2ï)

• 26
• =
(
(:>f(!1 1., Of) - ~1.oO;k" + l ';(l.,
li; + .'1) + &l.oli;)." )
-( :>f(g- /.,IÎi)- ~1.olÎ;.~-'" + ~'fl../.,Ii; + !/) + j!.oli;k,
-/.,0, - >!fl.oO,

(.\.:!!l)

whcrc,

1,3 1 '2 2;2.,fl


V3' = --2"("2°, (C3 - -'3) + 03) + (T(g-
.,fl·o 1·
+(T(L,Oj + g) + "21,2°')-'3 (,\.:10 )

L3 ··.,fl '., 1·


-03 - (-(g - 1,,0-) - -1,00')-':1
2 2 '2 •
.J2 1 00') 1"")
+( 2 \L,Oj + g) + "2L20ilc3

(·1.:12)

3V
3 F3 =
m 3'
m3'V32 J = (F3'F32 J
(
1n3v33 F33

• 2ï
• Th" f"rward n'cursior" for jiuk ·1 are il.' follow5:
(-1.:34 )

(4.35)

4 •
0
0) (~("
"f(
+ "~JO, + ~«, - "jO,O,)
"f( +
"-'4 =
(: 0
-1
1
o
C3 - 83)0 1 -
03
C3 83)0,03

( 4°,(0, h j + 4,,0,( <, - " j )


--
~ "f0IÔ3(C3 +
(4.36)
-"f0I(C3 - 83) 83)


whcrc,

• 28
• 1'·13 =
. d.l ".," "v~ '., l '.,
-d.,03 - 2 °i(c" - .-,,) + (T(.'/- 1"Oi) - :/oOi),-"
v'2 '., l"
-( T(L,Oi +.'/) + ~LoOih (·\.·Ill)

('1.:\:1)

• V"c3 (~,4,1

(,1,45)
)

(Il,46)

• 29

4.2 Bacl<ward Recursion for the Calculation of the Forces
and Torques

For t1", shov,,1 with four links. w" can compute: i: 5--+1[12J:

if.1 -_ i+IRi+l(.
i 1+1
+i F.1 (4..1ï)

i ni = iN i +'i+1 [,i+1 ni+l


1.
+ ip ci x iF i + ip i+l X 'Ri+1[
i+1· i+l (·1.48)

If link i is translationa!. then


. T·· .
" = Ifi 'Zi + b,di (4..19)

If link i is rotational, then

-.
'1-
- in·TiZ·.
1 1
+ b·O··
Il (4.50)

where bi is the appropriate viscous damping coefficient. Backward recursions for link
4 are as follows:

• (4.51)

·n. ~ 'N'{+ (~H::::) {) {:')


N.tln~,~m~v~<: ) m4(V~~I~) 2 /53

= N.12 + ~m~v'l<1 + d.i/Sl = n42 (4.52)


(
N43 n43
As Iink 4 has a prismatic joint, we have:

(4.53)

Backward rccursions for link 3 arc as follows:

o
o
0)
-1
(/51 + m 4<1) 3
4V
m4v4<2 + F3

• 1 0

30
/53 + m4v4<3
fil) (Ill)
• =
(
J", + 111"",,,,\ +
- f;.:, - l11.II'.k;t.+

m.II',Ic'.! + /. :r..\
f\"1. = .f:\'~

.f:\:\

(,U,(i)

(4,5S)

• 31

A, li"k :! i, fix('d. t1"'r<·for<·.
T{, = 0 (·1.59 )

Backward r('cur,io", for ii"k 1 are a.' fol\ow,:

_:11

~1 (h'JooJ::J,
:11 .,
1f
1 =
(~ 0
:11
)
+ (0)
0

=
C 0

1flj,.-J,,) )
-123 = (\
m,y

J'2 ) (4.60)
1(12, + 122) + m,y J'3


(4.61 )

(4.62)

The Td\, Td3, and Td.• are the ioad torques which act on the swing, hoist, and
crowd motors respectively.

• 32
• Chapter 5

Analytical Actuator Dynamics and Shovel


Parameters

This research proJect received excellent. collaborat.ion from t.he 1\'d 1 coml"UIY. It.s
chief mechanical and clectrical engineers met our t.eam members and l'rO\·id,'d a lot.
of data. such as various mass, size of machine. and DC mot.ors' current.s and volt.ages
and so forth. In the following sections, these dat.a will be used to calculat.l' param<'l.crs
of the analytical mode!.

• 5.1 Analytical Actuator Dynamics

The swing, hoist, and crowd motors on the type of c1ectric shovcl being modcllcd
are ail DC shunt motors. The circuit diagram for the DC shunt motor is shown in
Figure 5.1.

L
- il

~
Td
Rr + v(t)
ir
Lr

Figure 5.1: Circuit Diagram for DC Shunt Motor

• The notations in Figure 5.1 have the following physical significancc:

33

v( 1.) arnlatllfP \'oltag('

11 total Cllff('lIt

l,] fidd w;"di"g i"ductance

Il] fidd winding resis1.ance

i] --- field current

L" -- armature winding inductance

Il,, - arm..ture winJillg resistance

i. - armature current

E" - back emf

o- rotor position( radians)


w -- rot,ational velocity

Tm --- generated torque

T.l - load torque

1> - magnetic flux due to stator

The lumped modcl of a single link with actuatorfgear train is illustrated Fig-
ure 5.2. The equation for the armature current is:

V -E'•=
di. R,.'t
L• -+ (5.1)
dt •

The torque balancing equation is:

(5.2)

where b is the mechanical damping constant. Since the flux <1> is constant, the torque
deve10ped by the motor is:

• 34
(5.3)
• where kt is the torque coustant. 'l'hl' hack l'll1r n-Ial iou is

Ea = kl,o..· = k~~' = l''~TI


..
dO

where k 2 is the back emr constant. \\'I1<'n using; a g;l'arhnx wil h g;<'ar!lllx ratin r.
l ',..\)
1\1<'

(I.)=X2
ta = x J
vIt) + lIRa I--~-l _1_1-_ _
or S + 1 S
Ea-

"21----------J

Figure 5.2: OC Shunt Motor System Block

sum of the aetuator and gearbox inertias is as follows:

(;'.;' )

• From equations (5.2) and (5.3), we have

I m ,pO
dt 2
Considering (5.1) and (5.4), we have
dO
+ b dt = T,m - T.
r d = k'
'1 ta -
T.
r d (;'.li)

dO dia .
V - k'2 dt = Ladi: + Rata (;'.i)

Taking Tl = ~, T2 = 19'-, (5.6) and (5. i) become


,pO dO kl . rT,
• dt 2 + -dt = -bt a - -b d
To-

V k2 dO dia.
- - - - =TI-+ta (;'.9 )
Ra Ra dt dt
Taking Xl = 0, X2 = W, Xa = ia , and arranging the formulas (5.8) and (;'.9), we have
the state space rcprcsentation of the OC motor:

Xl = X2 (5.JO)
. 1 kl r
X2 = --X2
T2
+ -bT2
Xa - - Td
bT2
(5.11 )

. k2 1
+ - 1- v


Xa = ---X2 - -Xa (5.12)
RaTI TI RaTI

35
• 5.2 De Motor Parameters

1. CalculêLt.illg arrnatufe windiIlg rcsist.a.llce~ Ru:

Consid"ring "'Ination (5.1).

1,' - l;"[1, cii" + R'


= 1. u-/- al-a (5.13)
d

I;Vhcn thc rnotor is stallcd. E" = O. L" = O.


v
R" = .,-- (5.14)
1"

From P & lI's data, we can gel.:

6001/
R" = 1450A = 0.41 n (swing molor) (5.15)
550\1
Ru = 2000A = 0.28 n (hoisl mol or) (5.16)

• P... =
55O\i
-0'3 9 r>
1ï19A - .' - "

2. Calcn!ating the torque constant, k l :


(crowd molor) (.S.li)

From cquation (5.3), wc have

(.').18)

(5.19)

From P&H's data, wc can get(the field eurrent If = 100 A.)


2460
kt = ft.lb·f·4. (swing motor) (5.20)
1400
9000
kl = ft.Lb·fA. (hoist motor) (5.21)
2000
4i50
kl - 1125
ft.lb·f·4. (crowd motor) (5.22)

• Seeause

36
• 1 ft. = :lD.·IS

1 lb. = 0J154 kg.

1 kg. = 9.S N.
CIll. = 0.:lD'18 111.

thereforc.

2460 x 0.:3048 x 0J15'1 x 9.S ,.


kt = 1'100 = 2.:1~:l ;\.1II./:\.
9000 x 0.:3048 x 0J1.5·1 x 9.8 . 0'3' /
kt = - 2000 =b.l.;\.III.:\. (hoisl lII%r) (il.2·1)
4750 X 0.:3048 x 0J1.54 x 9."
kt = ----...,..,.-::-::----- = 5.726 N.III./:\.
1125

3. Calculating the back emf constant. k2 :

From equation (5.1) and (.5J1), considering very slllaH l,,,. wc hav,'

(.") .....')(')
)

• Thus
E" = k b<;"Jo'
"'·· = k·.w

dO
= k-,-
• dt = k·.w

From P&H's data, one can get(the field current 1/ = 100 A.)

275 - 0.41 x 600


k2 = = 0.27 v.scc. (swing motm') (",J .....
')lJ)
108.85
550 - 0.28 x 1000
k2 - = 3.44 v.sec. (!wist motor) (5.:30)
78.5
.550 - 0.49 x 1000
k2 - 87.92
= 0.68 v.sec. (crowd motor) (5.:11 )

4. Calculating the inertia on the OC motor shart, lm:

( :j 'J'))
" ••) -

(1) Inertia of Swing Motor, lm:

• 37

Frutll PkIr~s data. (me cau gd.:

1" = 100 "'./1. 2 = 100 x 0.·1.5'1 x 0.:lQ·18


2
= ·1.22 kg.m. 2 (.5.:3:3)
(5.a4 )
·1.22 + 20 = 2'1.22 kg.m. 2 (.5.:35)

(2) 11Icrt.ia of lIoist. ~Iot.or, 1",:

From P&Irs dat.a, onc can gel.:

1" - 925 Ib.ft. 2 = 92.5 x 0,4.54 X 0.a048 2 = :39.02 kg.m. 2 (5.a6)


\IR" __ ')00 (-6 .) -, 10-')"
• • = '>0'_).:"_) k
19 =
~ - X;) X _ ••)., X
_ ) 'g.m.-., (.5.a7)
2 2
lm = :,!J.02 + 202.a2 = 241.:34 kg.m. 2 (5.a8)

(a) Incrtia of Crowd Motor, /"':

From P&B's data, one can get:

• 1"
/9

lm
=
=
=
29.5 Ib.ft. 2 = 295">( 0.454
1\1 R = 100 X (20..50 X 2.54 x 10-
2
2

2
12.4:1 + la.56 = 26 kg.m. 2
X 0.a048 2 = 12.44 kg.m. 2
2
)2 = la.56 kg.m. 2
(5.39)

(5.40)
(5.41 )

5. Friction Coefficients of OC Motor:

The clectrical damping constant be can be expressed[66]


b _ k,k 2
e - Ra (5.42)

The rnechanical damping constant b can be expressed[66]


p
b=-;;-be (5.43)
w-
whcre P is the power of the OC motor.

(1) Friction coefficient of swing motor:


->"3 x 0 .)-(
k k = .)_..,<.>
= 2.2. .- = 1.57


be N.m.sec. (5.44)
R. 0.41

38
• b
__ 2iOO x la"
---,--:-:-:-=-
1OS.S5~

(2) Friction coefficient of hois! motar:


- 1.,:') 'j = 226.:n .Y.11l ..'" n°.

k,k, = 6.10:3 x :U'I


b
<
= Ra
---=--:-:--
0.28
= i·1.9S (:I.·Hi)

b=
·roo
-, X 10 3 -,. • -
_, _" - .'1.9:s = .36.3.\1 (:•.·Ii)
1 :S••)-

(:~) Friction coefficient of cro",d motor:

_ k l k 2 _ 5.i26 X 0.68 _ l') \-


b< - R - 0 '3') - _. 1 N.m.scc.
a .' ...

2iOO X 103
b= 8-1. 9')2
_ - _. 1-
1·) '3'>-
1 = .. >1. l')
- !'V .m..'cc. (;;,49)

• 6. Time Constant,

Taking L. = 0.01,
Tl:

La 0.01
Tl = -Il. =--
0.41
= 0.024 sec. (swing molor)

TI = L. = 0.01 = 0.036 sec. (/wisl mol or) (;;.;; 1)


R. 0.28

L. 0.01
TI = Il. = 0.32 = 0.031 sec. (crowd molor)

i. Time Constant, T2:

lm 24.22 _
T2 = b = 226.31 = 0.101 sec. (swing molor) (5.53)

lm 241.34 _

• = b = 36:3.1i = 0.61
T2 sec. (hoisl molor)

39
• K
fT"
,~ = -, = 'Ir l') = (J.Oïï
)

(;(,.t1' 1J{)x ratio. r:

l' =
••

1
:W
j. -

= 0.002(;
(rroll'<I

("lI'i1l'l :110101')
111 01 or )

(5.56)
:1~2.:1G92

1
l' = = 0.02 (hoi.,1 1110101')
'lï.8G4ï
1
l' = = 0.02,1 (CI'oll'd mol 01') (.5.58)
,12.151:1

5.3 Rigid Body Mechanical Parameters

From P&H's data, wc can get:

LI - 1:32.00 in.= 11 JI. = :3.4528 m. (5..59)

=
111.1 341000 lb. = 1.54814 kg. (ganll'Y nlass) (5.60)

• L2 =
n~2

L3
-
=
=
240.00 in. = 20 JI. = 6.096 m.
135000 lb. = 61290 kg. (boom mass)
250.00 in. - 240.00 in. = 10 lTl.
0.833 JI. = 0.2539 m.
(5.61 )
(5.62)

(5.63)
m3 - 100 lb. = 45.4 kg. (saddle black '7lass) (5.64)

L.I = 354••'J-'
_il lTl. - ')9 -.)
_ .il_ Jt • -
- 8.96 m. (5.65)

rn'l - 44.000 lb. = 199i6 kg. (dipper handle mass) (5.66)


m5 = 8iOOO lb. = 39498 kg. (dipper mass\ (5.6i)

where the physical significance of LI. L 2 , L 3 , and L'I are shown in Figure 2.1.

5.4 Rigid Body Inertias

Inertia of the rectangular prism is (sec Figure 5.3):


m
-lb"
12
+ c·)
0 0
(5.68)

40

III ., .,
/
''''
= -10'
1~
+ co).
111 .• .,
/.... = -(0-
1~ -"-, b-).

1. Gantry Frame Inertia

The gantry franH' is a fectanglllar prisnl~ t.h('refon' \"l' h't\"(,.


= /7/1(r.
2
+ c2 )
15·181·1((:\00 x ~f + (~OO X Il:;';'')2)
_

12 - 12
15'1814 x (58.06 + 25.81) 1."
= 12 = 1082021 "fI· III .- l'i.71)

2. Boom Inertias \Ve can approximatcly cOllsider t.he boom as a n·l"t."II~III"r prislll.

T IL----I-l--- x
o
1.. . . .;.-.__.._._--
b •

• 1-1 --.---1

Figure 5.:3: Rectangular Prism

Therefore, we have:

I rr2 = ..
( .).1-'))
...

l yy2 - (5.7:1)

1::2 - J"y.TTl. 2 (5.7'1)

where a = 600 in. = 15.17 m., b = 1 m., c = 1 m.

3_ Saddle Block Inertias

The saddle block has a complex trapezoidal shape. We use the fUlldament.al
inertia calculation formulae to solve the problem.

• 41
• 0" t.he x-y pla"e (sec l'igure 5..1), t.he right. si de of t.he lrapezoid has:
(.5.i.5)

IV-a y-a
= (.5. i6)

(.5.ii)

The left side of t.he trapezoid has:

w-O y-D
- - =x-a
b-a
-- (5.i8)
10('" - a)
y= (5.i9)
b-a

Since rn = (a + b)whp (p is materiai density), therefore,


x


Figure 5.4: Saddle Dlock

= rh rW(z2W(:r - a) _ z2w(x+a) + ~(w(x -a)? _ ~(w(x+a))3)pdxdz


Jo Jo b- a a- b 3 b- a 3 a- b
2
- 2i.33 kg.m. (5.80)

(5.81)

• 42
• = 21 ï5.:3'1 kg.111.~

According 1.0 the centre m",," formula for a t.rapezoid.


( ." S")
J.~ _

(',) (~)
.r e :~(ll+b)

y, = 0 - .... )
( :).~~.,)

=c a
In terms of the parallcl axis theorem.

C Il
.r:3 = r.r3 - rn3 (Yc ~ + "'c• --
~) 1r.r:\ = .)- .,., k"!I.1U. .,
... 1.'>.,) ([>.S·I)

(.'i.S5)

• In aIl the above calculations, we choose:


(.'i.86)

m3 = 100 kg., a = 0.5 m., b = 1.0 m. (5.8i)

w - 1.5 m., h = 0.3 m. (5.88)

4. Dipper Handle and Dipper Inertias

We l'an separate a rigid body into two parts: dipper handle and dipper. Each
part l'an be consiJered as an appwximatc a rectangular prism.

(1) Handle:

a - 354.25 in. = 8.96 m. (5.89)

b - 41.00 in. = 1.03i m. (5.90)

• c = OAm.

43
(5.91 )

l 'J97(j,
=
III,
-(s.!Jrj·
" "
+ (1.,1-) = --(S.!J(j- + IH-), "
= 1:1:190:;.79 1.-.'1.111.- (:;.92)
12' 12
TfI.J -, ., 1!J9ïfi .,., ,.,
1~/'J'II = -(s.!Jr;- + I.lJ.l") = --(S.!J(j- + 1.01-) = 1:1:;·129.:;7 l.y.III.-(:;.!):I)
12 12
III" ,_, " 1997(;. '. __ ,_, "
= -(1.0.l,- + 0.·1-) = --lO.Ir..,. 1.0,:») = 20.')6..1~ l.'fJ.III.- (:;.9,1)
12 12

(2) Dippel':

Il = ~·17 i1l. = :J.72111. (:;.95)


b = 117 in. = 2.96711. (5.96)

c = 1·17 i". = :l.72 m. (5.97)

= m'(;l.72~ + :l.72~) = :>9498 X 2 x :l.72 2 = 91098.19 kg.m. 2 (5.98)


12 12
= -m'('l-')"
12
.. ,.- + ,)_.""6")- = - - .. lS·'1 + lS.I
:39498(I'l
12
'-6) = -1'39'31-
/" '. / k-g.m. 2 (-;). 99)
12 •.96~ +,'3 ./-


- m,(.) -.)2)
= 12 3 "1) = -1'3'-
39498(,lS./-6 + l ..lS· /" lS/. 9 k g.m.;).
2 (- 100)

The distance betwccn the mass centre point of the combined hody (i.e. dipper
hamlle and dipper togcther) and the mass centre point of the dipi>er handle is Dl; the
distance betwccn the mass centre point of the combined body and the mass centre
point of the dipper is D2 • Thus,

(:::) = U.l (5.101)

ln tenus of the parallc1 axis thcorem,


(:::) = rD (5.102)

• CIr.r·J 1 =
44

"1 !J.u·l 1 = 1!Wl 1 - III., (( .r,. ').,- + ( Z., 1)·')
-

= laO·'l:35.57 IN/.IIl.:! ('">.\ 01)

CI ::,1 1 -1
- ==., 1 - 111.1 ((..r.. I)~+( !I.. I)~) -1 - _ .) ).' I~
- =:.1 1 -"0"(' /,.
'~!I.lIl."

= 91098.19 - :39498 x o.:f = Sï;;:l:t:Jï kg.HI.:! (,>.1 ()(j)


= 74:39:1.17 - :)9·198 x 0.a2 = 708:38.:)5 kg.nl.:! (ii.IOi)

CI::.," = I::.,~ - n/5((x/f + (y/f)


= I::.,~ = i4a9:Ui kg.m.~ (:>.I()~)

CI.l/" = CI...,' + cl..l = 1:30·1:3.5.5i + iDS:18.a5


_ 2012ia.92 kg.m. 2 (5.110)

= i64t.9.65 kg.m. 2 (5.111)

At this point, we must note that these analytical calculation of inertias arc


approximate. The precise results must come from actually measuring and estirnating.

4.5
• Chapter 6

Friction and Impact between Bucket and


Muckpile

When the shovP! is digging in the muckpile and loading rock, a digging interaction
Ill't.wcen the bncket. and the rock exists and affects the overall machine motion and
performance. Therefore, this interaction is an important. issue ",hich must be ad-
dressed by any modcl of shovel operation. The interaction actually inc1udes two
aspects. friction and impact. This interact.ion has not becn studied extensivcly by


anybody t.o date. Rowlands, (1991), however, has studied the friction of dragline
buckcts, and we can obtain sorne insight from his work. Based on his work, we have
devcloped these friction models into a general coordinate space representation. In
addition, wc have also studied the impact effects relevant 1.0 larger fragments of rock.

6.1 Friction

Whcn the bucket digs in the muckpile, the torque acting on the shovel bucket is zero;
the forces acting on the shovel are shown in Figure 6.1. The physical significance of
these notations are as follows:

P - payload weight

T - tooth friction force

L- lip friction force

• f - friction forccs; this inc1udes friction of the bucket bottom's outer surfaceUd;

46

,.

Figure 6.1: Forces Acting on Sho\,<·1 l\'1aehin,-

friction of the bucket bottom's inner surface(f2): the outer ",rfae,' frict.ion ,,1' I.wu
bucket sides(fJ); the inner surface friction of two bueket. sidcs(f,).

1. Frictions on the Bucket Bottom's Outer and Inner Surfaces

• One can assume that the normal ioad increases al. a constant rate dlll' t.o 1.1",
payload. The friction of the bucket bottom 's outer surface(f,) can he expressl'd hy:

where A-h is bucket mass; Mp is final payload; 0 is ma:..:imulll digging distance;


JLlx~ + Lly~ + Llz~ is aetual digging distance in work location coordinat.. {6}. (Llx" =
X6l -X60, LlY6 = Y61-Y6O, Llz6 = Z6'-Z6O), (X6'YG,Z6,) is destination of buckct digging;
(X6OY6OZ6O) is start of bucket digging; 'P is digging angle; Il is coefficient of friction.

The friction of the buckct bottom 's inuer surface(f2) can be exprcssed 'L' follows:
(6.2)

where Dl is the bucket length.

• 4ï
• 2. Friction on two Bucket Sicles' Outer ancl Inoer Surfaces

'l'Ill' ""t"r a"d i",,,'r sllrfaCt' friction of 1\\'0 bllàet sides (f,.f.,) can be described
li:> 1Ill' passi\'(' tlirllst 1'''1' Illlit lengtli along a stirr rd.aining wal! [iD]:

(6.:3)

wlien'. l'l' is lateraltlirllst pel' lengtli of wall. -1 is unit weight of the iiI! material.
II is actllal d"ptli of material. and k l , is passive pressure coefficient. The outer surface
fridioll of Lwo hncket sides. h. is:

(6.4)

where III" is tlie linal average depth of material in Lhe bucket, n is length
number of bucket.


The inner surface friction of the two bucket sides, f." is:

(6.5 )

3. Friction on Teeth and Lips

The friction across the tccth and lips of the bucket is a very complex problem.
A modcl [:30] was proposed by Hettiaratchi and Rccch as follows:

(6.6)

where R is resistive force on blade, X is blade tip depth, w is blade width, and kr is
a factor duc to gravity derived from nomograms. Thus, one can obtain teeth force as
follows:
T(~X6, ~Y6, ~=6)
IiI ")2 (~x62 + ~Y62 + ~Z6)P.
= nnwTk D r (
2
(6.ï)

Similarly, IiI' force is:


(6.8)

48
• \\"here /II is number of lel'th
and lips respecliyely,

By
l>r lips and 1/'1' and /l'l, r"pr""'111 Ih,' \\"idth ,,1' th,' I""th

and large. in t.he abo\"(-' thrce parts of frict.ion. ~(}nlt' paratlH'tel's al"t' ClH\:-:tatll.

(c.g. Ah D h ctc.). and SOI ne paraIllctC'fS call be oht.aitll"<t by l'll~iIH'('rill~ IlliUlllal ('.~.

k p and k r ), As soon as \\"e kno\\" ~Xl;' ~Yti. and ~=l; \'ia l',,r\\"ard kitl<'m;üics ,,1' I.h"
shoycl, equations 2.25. 2,26. and 2,2i. \\"e l'ail cakulat" Ih",,' l'riclions, Th" r"I,,\'allt
simulating \\"ork has becn induded in our simulator.

Rowlands used simi!ar formulae 10 cakulate his fricl.i,," mo("'ls for drap;lin"
bucket, but he omitted the friction coefficient. l'. in his formnla". 'l'h''l'<ofon·. his
models arc unreasonable because of the different physica! Illeanillg on b"t.lt sid,'s of
his formulae. In addition, he only considered the one dinlt'nsion l'a.''': 11<'1"'. Ihn'('
dimensions based on real wor!d coordinates arc introdnced.


6.2 Impact

During the digging process, the shovcl often mects with large fragmellts which ('nl."r
and interact with the bucket in a style different from smaller fragments: they prodllC<'
impact on the bucket and the whole machine. The impacts acting on t.he shov'" are
shown in Figure 6.2.

In terms of the laws of motion, the impact bet.wecn two collid; Ilg bodies is:

(6.!))

where 1 represents bucket and 2 represents fragment. Thercfore, MI is n"L'S of


bucket; VI is velocity of bucket; M2 is mass of fragment; V2 is vdocity of fragment.,
We can define colliding compression, T. It should have:

(6.10)

When two bodies collide, the deformation force is expressed by the Hert.z Law [20]:

• 49
(6.11 )

Figure 6.2: Impacts Acting on Shovcl Machine

where K2 = 3(.'H;'~,+Q2' S, + S2 arc material-dependent constants and QI,


(d2, and Q3 arc geomctry constants. Differentiating equation (6.10), and considering
equations (6.9) and (6.11), we gct:

• where K, = A~, + A~2'


int.egration or eqnation (6.12) is:
t = _ F _ F = -K t KoT3{2
M, M2 -

When two bodies begin to collide, T = 0,


(6.12)

l' = \Ii. The

1 ?
:)(1'2 _ \li 2 ) = -":Ç;J{,J(2 T 5/ 2 (6.1:3)
_ .J

When a collision is completed, the relative vclocity l' = 0, and the compression
has a maximum value. From equation (6.13), we have:

(6.14)

Substituting (6.14) into (6.11), we have ma.'(imum impact:

F = J{o( 5\1? )3/5


- 41\1'\2
l' l'
~ 7) V1 (6.15)

• WIlere TJ = l\2
l' ( 5
'IKIK~
)3/5 •

50
• FraIll cquatioll (:L2·1). we ("an gl't. tht' itllpact alntl~ 11Il' .r;. élud :-~l dircctil\Il:". i.t'..

FI
. .
= (.1'1,11, + J"IO{l" + .1"".!'1 )11
.
(lU li)

((i. 1ïl

In short. in cqnation (6.15). w" know /,'0' .\1,: and \', <"an Il(' .."kn\al<"d I>y
the Shovcl Jacobian. Therefore. a$ soon a$ w" know th" ma~s ,'r dimension of 1\1<'
fragment (i.e. 1Hz ). we can calcnlate the impact.. NOl'mally. if Mo iIl<T,'a,,'s. h', will
decrease. and from eqnation (6.15). impact F will incn·a~". That nll'a,1S I.hat I:tl·~,,1'

fragments will have largcr impacts on t.h" shovcl. Th" rd"vant "<"<'<lItnt ha~ h""n
included in onr simuiator.

6.3 AlI External Forces and Their Calculations

• l. AIl ExternaI Forces

We put ail externai forces together and gel:

ft (xG) - (lyh + '~Jxa + ya)Y/lcosr.p (G,I~)

M
Jz(xG) - D~ Jxa + yayIl cos r.p (G.I !J)

,K (H f ")2( Z
D XG+YG2) /"
p
!3(XG) - 2 (G.:W)

,K (Hf ")2( Z
D xG+ YGZ) Il
p
!4(XG) - --:;- (6.21 )
- 1
Hf" 2 2
- n,wr kr( D) (x G+ YG)1l
2 (e) ........
'N)
T(xG)
, Hf" 2 Z
n,wI.l\r( D) (X G+ YG)1l
2
L(X6) - (6.2:1)

FI - (XIlOI + XI203 + X13 d4)I) (6.21)

Fil = (X3'O, + X3Z03 + x33d4)I) (6.25)

The payload gravity can be divided into:

• fp = f pr; + f p ,;
51
(6.26)
• ThIlS, Liu: lotal force acting on the hucket is:
(6.27)

(6.28 )

(6.29)

(6.:30)

(6.31 )

The scalétr valucs arc:

(6.32)

(6.33)

If there arc no large fragments of rock, FI = Fu = O.

2. Calculations of Ali Forces

• According to P&ll's data, we have:

.fI,h - 87000 Lb = :39498 kg, lHp = 10000 kg


II - 1 m, 12=6
cp = 20·, kp = 4..5 (6.34)

i = 1.5500 N/m 3 , tuT = 0.42 m

tuL = 0.:35 m

Thereforc, for a complete digging proccss, the final force eifects on the bucket are:

/1 - (39498 + 10000) x 9.8 x 0.58 x 0.939 = 264184.5 N. (6.35)

h - 10000 x 9.8 x 0.58 x 0.939 = 53372.76 N. (6.36)


15500 x 4.5
/3 = x 12 x 0.58 = 20227.5 N. (6.37)
2
15500 x 4.5
f, = x 12 x 0.58 = 20227.5 N. (6.38)
2


T = 6 x 15500 x 0.42 x 4.5 x 12 = 175770 N. (6.39)

-.)
0-
• f.

J,.r
Jp , =
Fra =
=

=
G x 15;;00 x 0.:3:') x .1." "' 1" = l·\(i·li;; .Y.

10000 x 9.8 x O.:~·I:! = :1:1:)1 ï.9 .Y.

tOOOo x 9.8 x 0.9:19i = !):!08!l.!l .Y.

-:!G·118;1.5 - 5:ni2.iG - 20:!:!i.:) - :!():!:!i.;;


lli.IO)

l (i.·11 )
l li.·I:!)

1i.5ïiO - 1;IG;li5 - ;;:351i.9 = -il:liï;;.IG S. (li.·I:1)

F,s = 92089.9 ;'Ii. ((i.·I·1 )

In case of impact, we assume FI = !JSOO NO'. Fu = ·I!JOO N.. t.1"'1l add t.hese \'allles
into Fxs and F,s. Inside our simulat.or.

(G.·\(i)

where JSl and JS3 are the force clements acting Oll the shove\. These el'temal forcl's
have been included in our simulator.

• ,53
• Chapter 7

Data Acquisition and System Identifications of


Actuator Parameters

ln t.his chapt.cr, wc will ïocus on t.hrcc actuat.or models (swin!! mot.or, hoist motor,
and crowd motor), They are one of our main research objectives, because the rigid
body dynalllic and kinematic crfects, digging process influence, and the control system
design ail relate directly to thcse models, The work consists of t.hree parts: the first is
d«ta acqnisition: the second is analysing thrcc transfer function models by using the
ARX lIlct.hod; the thir<! is setting up state space models by using the MLE method,

• The software MATLAB is used in system identification.

7.1 Overview of Data Acquisition

The field test," were undertaken 1.0 acquire data for use in system identification and
subsequent validation of the shovel simu!ator. The data acquisition \Vas condueted
al. a Western Canadian coal mine. The tests involved monitoring and data logging
using a PC-based data acquisition system. The data sampHng rate was 100 Hz. The
b'L.~ic structure of the data acquisition system is shown in Figure Î.l.

7.2 System Identifications of Transfer Functions

'1'0 gct precise actuator modds is the main advantage of the system identification of
transfer functions. Typicall..·. we use ARX and ARMAX mode! identification meth-

• ods. They are stated as follows:

54
• AID --
;lml;llun: curn:nt

aml;UUrC \'Olt;lgC

ann cXlcn:-ion
Crowd
l\'~()tor

C
0 annalurc currcnt

N Hoist
PC armalUre VOIt'lgC SHOVEL
V
Ponable E fOpe position Motor
R
T
E
R l arma. vol.
Swing
angle
Motor

Figure i.l: Data Acquisition System for Shoyel Machin!'


If the system has input u, output y. and oulsicle disturbancc !', t.\", re!at.ionship
can be wri 1. ten

y(t) = G(q)u(t) + ll(q)c(l) (i.! )

where q is the shift operator and G(q)u( 1.) should be:


«>
G(q)u(t) = L g(k)u(l - k) ( -, '-1
'JI
k=1

and
00

G(q) = L g(k)q-k (i.:\)


k=1

e(t) is white noise with variance>' and <Pv(w) = >'jll(é"W. The above systelll can be
reprcsented by a general parametric mode!.
B(q) C(q)
.4(q)y(t) = F(q)u(t -nk) + D(q)c(t) (iA)

The general mode! structure can be divided into concrete rnodels such as ARX
and ARMAX. The ARX model is:
_nkB(q)


G()
q =q -- (7.5)
". .4(q)

55

1
1/('1) = - - (i.6)
..\('1)

will'I'(' B aIld A are poly:u)miais in the dday openltor '1- 1:

o· \()
<; = ,
l Tlll'! -) -'-, '
jllnr:fJ
-M
(i. i)

if = l)1 -r-'6 '2.'/ -, ,


I;)>() T'6 nbq -,,'+' (i.S)

wl",re t.he lUI alld n6 arc t.he order of t.he po!y"omials. The nk is t.he dclay from input
t.0 Ollt.pllt. \Vc cali write t.he modcl t.o a standard form

A(q)y(1) = 8(q)u(1 - Ill,) + c(1) (i.9)

The ARMAX modcl is

-"k B(q)
C()
'1 ='1 -- (7.10)
.4.( '1)
H( ) = C(q)


(ï.l1 )
'1 A( '1)

'l'hus, the model has a structure

A(q)y(t) = B(q)u(t - nk) + C(q)c(t) (7.12)

where 04('1) alld B(q) have same meaning as AR.X mode!, while

(7.13)

ln t.his section, wc mahly used the ARX method, while the ARMAX methocl
was uscd as a refcrcnce. Thr. typical methods and steps for the system identification
of the transfcr function in our rcsearch are as follows:

Subtract sampie meaas from acquired data; compute loss functions for a set of
dirrercnt model structures and obtain the pole-zero plot; select the mode! order (struc-
ture); use the ARX or ARMAX method; compare l'cal data 1.0 model output data; if
not satisfactory, try again and unti! satisfactory. The relevant system identification

• procedures of the transfer function are shown in Figure 7.2.

56
• Stan
ldenliftcatio

Sublmcl
Sample Means

t
Compule
Loss Funclion

~
Oblain Pole-Zero
PlOI


Selecl
ModelOrder

Use
ARX or ARMAX

Compare
DaIa

~
No Satisfactory ~
? -~

Figure i.2: System Identification Procedure of Transfer FUllctions

• 5i

III OUf (·xJH'rirrwtl!.. input dat.a carne from an operator"s input and wc cannat

p",du('(' hil;h/low f,e'lu('ncy signais in input data. The,efo,e. the input-output data
('ail lw 111lSuitahh: h(~C"éLllS(' therc is Ilot enough infortllatioll. typically duc ta lack of

fn'qllf'flCY COllt{~I1t. \Vhen wc: do s~!stcm identification ll~ing the dat.a. our c~tinlation

can still convcrge 1.0 a solution - modeL i.e. modcl simulated output can match actual
output, but t.his modcl docs not match the t.rue system pararneters. In order to obtain
IUOr<' accurat.e modcl which can match truc syst.em paramcters, wc even need design
Ollr experilllents again before processing the procedure of system identification in
Figure Î.2.

(1) Swing Motor:

According t.o t.he procedure m Figure ï.2, the t.ransfer function of the swing
motor is ident.ified as follows:

At. first, wc choose voltage as input, swing motor angle as output, and remove
t.he sam l'le mean. Then we compute loss functions and show the pole-zero plot. This

• is the key step in deciding the order of the modeL The pole-zero plot of the swing
mot.or is shown in Figure ï.3.

3
OUTPUT' 1 INPUT' 1

,
2

-, o 2 3

Figure ï.3: Pole-Zero Plot of the Swing Motor

• 58
• AMPLITUDE PLOT, Input. 1 outpul • 1
10"rf- - - - - - - -

'o"

10'
Ircqucncy (rad/sec)
PHASE PLOT. input 1 1 oulpul # 1

-2000'-:------~:__-----.....,.-------'
,~ ,~' 1~ 1~
lroquoncy (rad/sec)

Figure i.4: Frequency Response Plot of the Swing Motor in Discrete-Timc

• 10'
AMPUTUOE PLOT. Input . , oulpUl' 1
r-----~-----r-----~----_,

10..' - : - - - - -.....;-----'";:------'-;,.-...:::::.....-......J
,~ ,~l 1~ la' 1~
Iroquoncy (rad/soc)
PHASE PLOT. Input' , output" ,
5'..00,.------..----...;.--.:....--.-----..,

s °

~
! -sooo

-10000L.:-----.....,.----"""::------'-;----...J
1~ 1~ ,~ ,~ ,~
trequency (rad/sec)

Figure i.5: Frequency Respcnse Plot of the Swing Motor in Continuous-Time

• .59

In Fi:,!;llrf' 7.:L url(' C;ltI :-;('(' that tlH'r(' ;In~ only three poles. Therf'fon~. the SYStCII1

lH'f"I)I1If'S third-ordf·f. :\rt.(~r l1siIl~ tIl{' ARX IllCt.hod. one obtains the transfer fUIlclion
of 1.1", swing Inol.or 'LS follows: (a) Discrete Time Transfer Function:
-0.00 Il = - 0.0024
G(=) = =-., - 0.OG9·1=-' - 0.OG99= - 0 .~'GO):. (i.14)

TI", frequency re'sponse plot. i.e. Bode Plot. in the discrete-time. is shown m
l'ï~lIre ïA.

(b) Continuous Time Transfer Function:


G s) = O.0004s 2 - 0.001Gs - 0.005.5
(i.15)
(. S3 + 0.150G s 2 + 4.:399i s + O.OOOS

The frequency rcsponse plot. i.e. Bode Plot. in continuous-time. is shown m


Figure 7..1.

(2) Iloisl. 1Vlol.or: Simi!arly. one can choose voltage as input. rope position as output.


OUTPUT. 1 INPUT.'
~--,

••
•.6



-o.•
-0.8 )(

-,!----;:;---=='''--:--O=::::--::7'----!
_1 -0.5 0 0.5

Figure i.G: Pole-Zero Plot of Hoist Motor

. and obtain the pole-zero plot of the hoist motor shown in Figure ï.6. One can find
that there are one pair of symmetric zeroes and one pair of symmetric l'oies. Both
pairs cao be dcleted. As a result. there are only three poles. Therefore, the system


becomes third-order. One cao obtain the transfer function of the hoist motor:

60
• M1PUTUOE PLOT. Input. 1 Qulpul • 1
lO"r!--------------_----~-.~. _•. -

10. 3
10':
Iroquoncy (md/soc)
PHASE PLOT. input 1 1 oulput , 1
200

'00
0
~

!
0

-100
'0" 10' 10'
lrequonc:y (l'3d/soc)

Figure ï.ï: Frequency Response Plot of Hoist Motor in DiscrctL~Tilll<'

• AMPLITUOE PLOT, input" 1 oulput " ,


10-'.------~-----~'------_ _- - - - - ,

'0"
10"

'O~'-:-----
,~
. . _ ,_-----'-:c-----'----"--...J
. .1~1 1~ la' 1~
lroquency (rad/sec)

2OO,-----r-----r-----r----
PHASE PLOT. input" 1 output. 1

~O~. : O·~,·- - - - - '...0 ':'·-----'0":'------',0'


. - - - - - ,..
trequency (rad/soc)

Figure ï.S: Frequency Rcsponse Plot of the Hoist Motor in Continuous-Tirnc

• 61

(,,) Dise,,·te Ti n'" Transfer Funet ion:

-0.00:'·1="
(;(=) = """'=:-,--0-.0-6.-1=--,"-=-.:..:..::.0..:..06.:..:·:3-9-=---0-.::;.,...'i-I...,.8 (i.16)

TI", fre'luene)' response plot. i.e. Bode Plot.. in diserete-time. is shawn in Figure i.i.

(b) Continuous Time Transfer Function:

r.(.) _ -O.OO-1ls" - 0.008-1.' - 0.008.5


lJ' oS - (i.1 i)
s" + 0.I:3i2s" + 4.:3968s + 0.000.5
The frc'lllene)' rcsponse plot, i.e. Bode Plot, in continuous-time, is shawn in Fig-
nre i.S.

(:1) Crowd Motor:

Similarly, one ean choose voltage as input, arm extension as output, and also
obtain the pole-zero plot of the crowd motor shown in Figure i.9.

OUTPUT' 1 INPUT' 1
15r-----~--~--~--~---,

• '0

o '-'-'-'-'_._._._._ ..c~._._._._._._._.~_._.

-5

-la

-15''::--~':--~--~--~--~--
-15 -'0 -5 a 5 la 15

Figure i.9: Pole-Zero Plot of Crowd Motor

• 62

10'
frcquoncy (md/sec)
PHASE PLOT. input # 1 oulpul • 1
l000,------~------~-----._.,

~ -'000
i
~
-2000~--------""",--"-----'
,~ 1~' 1~ la'
lroquoncy (md/soc)

Figure ï.lO: Frequency Rcsponsc Plot of Cro",d Motor in Discrdc-Tilllc

• 10'
AMPLITUDE PLOT. Input # 1 output" 1
r-----~-----~----~----_,

,o..'-:- .....;- '";;- """'-; -J

,~ ,~1 1~ 10' lif


Iroquoney (rad/soc)
PHASE PLOT, Input" , output # ,

-2'7----........-:-----'-:----.....,,------'
,~ 1~ ,~ 10' 1~
frequençy (radfsoc)

Figure ï.ll: Frequency Response Plot of Crowd Motor in Continuous-Timc

• 63

Ifi Fi~lln' ï.!J. ofle ~afl fifld th,,! there arc three poles. Therefore. the system is
third·order. :\fter tJsifl~ tlll: :\RX flwthod. one cafl obtain the transfer function of
tI", nowd lflotor 'L' follows:
(a) Discrete Time Transfer Function:

G(-) = 0.0002=2 - 0.00019= - 0.00.5


(ï.18)
- =3 _ 0.0168=2 - 0.0169= - 0.96.54

The frequeflcy response plot, i.e. Bode Plot, in discrete-time. is shown in Figure ï.10.

(b) Continuous Time Transfer Function:

-0.002ïs + o.oOïïs - 0.0099


2
Crs) =
(ï.19)
S3 + 0.03.52s 2 + 4.38ï3s + 0.0013

The frequency response plot, i.e. Bode Plot, in continuous-time, is shown in Fig-
ure ï.1 1.

The above motor frequency responses exhibit a resonant peak at 2 rad/sec.


Upon inspection of the experimental data, it was ddermined that this peak was an
artifact duc to system and measurement noise. Renee, an alternative technique for
dctcrmining the motor modeIs, which is not acceptable to such noise, must be found.
Therefore, wc chose the ma:·.:imun likelihood estimate while the models obtained by
using ARX method are used as a reference.

7.3 System Identifications of State Space Models

The maximun likelihood estimate is usually used to identify the state space model in
a\"ailable System Identification Toolbox of MATLAB. In addition, we use Simulink in
l\latlab as our simulation too1. In order to use the tool, we must know a variety of
parameters of state space model exactly for the DC motor. Usage of the simulation
tool will be covered further in Chapter 9.

Recall the typical DC motor's third order system, (5.9), (5.10), and (5.11) as


follows:

64
• .r, =
X2 =
.f;!

k, 1 1
l7.:!1l)

l7.:!l)

X3 = ----x.. - -.r:} + --1'


Rrl,t - TI Rn 71

The coefficient matrixes arc

- ')")
( 1 ..... )

C=(l 0 0) (7.2·1 )

One uses the Ma.'(imun Likclihood Estimation (l'vILE) for the iùenlilication. The
fundamental principle of the ma.'(imun likclihood estimation is: snppose that th,'

• observations are represented by the random variable z" = (z(l),z(2), .... z(N)) that
takes values in RN. Therefore, the probability density function is

f(O '~ .... 1 ~ ....2, ••• , "'jV


0 - - )
-
- f : (O', ...oN) - ')')
( 1 ••')

The probability is
(i.2G)

The purpose of the observation is in fact to eslimate the vector 0 using zN. This
is accomplished by an estimator Ô(zN). If zN is the actually observed value, the
resulting estimate is Ô* = Ô(zN *). There may be a variety of eslimators, i.e. Ô" Ô" ,
O~. Subrtituting these values into equation (i.25), we have f=(Ô t ; zN), f=(Ô,; zN), ,
f=(f)~; zN), etc. This function, !=(Ô,; zN), is called the likelihood function. Il rellects
the "likelihood" that observed events have taken place. A reasonable estimator of 0
should be selected so that the likelihood of the observed cvent becomes "as maximum
as possible", i.e.
ÔML(ZN*) = arg max!=(O;zN*) (i.2i)


.. or '
o

65

Titis is IIH' rJltlXilll1lI11 likdih()()r! ('st imator.

\"f)\\'. \\'1' Ilf',I!)t1 ft) c'sliIllalp thoS(' paratllet('r~ wilirll rwed 1,(1)(' id(,Ilt.ified, Itl

:"'rms or (ï.:!:~). \'0(' a~Sl1l1l(' some c()(~ffici('Il1, tnalrix(~s ët..'" rollows:

Na.\'
NaN
N:N)""
NaN
( o
0 ) ./\ =
NaN
(N;" ) (i.28)

Theil. according 1,0 the parameters offered by P&H Ltd, wc assume these initial
vailles:

guu:....... -- ( -_,lu
')-Q
G -:jQ -:lQ 5) : (i.29)

Meanwhile, wc define a state-spacc structure with unknown clements. Before


wc use the maximum likclihood mcthod, wc package standard state-spacc parame-
t.erizat.iolls into the Illda modcl format (MATLAB's structure). Wc obtain detailed
paralllc\el" values lIsing t.he maximum likclihood method. Finally, the est;mated val-


11<OS will be cOlllpared with act.ual values. If the residuals arc 1,00 big, wc can renew
tll<' yu""" value and repeat the above steps. After identifying thrcc actuators, wc can
gd their st.ate space parameters as follows:

( 1) Swing Motor:

~ = 14.2.542 (i.30)
'2
kl
bT2
= 6.6362 (i.31 )
ko
--- = 32.5403 (i.32)
Raïl
1
- = 31.3358 (i.33)
1
-R = 11.9410 (i.34)
aTI

From (i.30). wc get T2 = O.Oi. From (i.33), wc gct TI = 0.0319. From (i.34), wc gct
1
Ra = = 2.6242 (i.35)


0.0319 X 11.941

66
• 10
,
l ~--
AMPLITUDE PLOT. mpullf 1 oulput JJ 1
_o. -.... "1

10~

10~
Ircqucncy (r.ld!SC'C)
PHASE PLOT. input # 1 oulpullt 1

o
-1
~
i -100

-200'-;;------'-;------'-:;---------'
1~ 10' 1~ 1~
Ircqucncy (md/soc)

Figure ï.12: Frequency Response Plot of the Swing Motor in Discrl'l.e-'1'inH'

• la'
AMPLITUDE PLOT. input # 1 output" 1

10. 10
la' 10' 10' 10' 10'
lroquoncy (rad/soc)
P~ASE PLOT. input ft 1 oulput It 1
100

0
~
0
i

_,oo~ _____'_,;__"__-'-;--=::::===;==:...-.-J
,~ ,~ ,i 1~ ,~
froquoncy (rad/soc)

Figure ï.13: Frequency Response Plot of the Swing Motor in Continno1l1s-'1'in",

• 6ï

Sllh~tittlting l7_:~;;) into \7'.:t~L wc ~ct

B('rau~c 7'~ = f- wC' han'


1 :21.:2:2
b = - = -- = :;·ll;
~~ O.Oi
Substituting (i.:3i) into (i.:\1). we gel

( -• ,.h
"S)

The frequency response plot. i.e. Bode 1'101. in the discre!.l'-t.illl'·. is shuwn ill
Figure i.12. The frequency response plot. i.e. Bode Plot. in cOIlt.inlllHls-tilll". is shuwlI
in Figure j.13. Comparing the two figures with Figure jA and Figure j.;;. we l'an liud
that there are no rcsClnance (at 2 rad/sec) in Figure j.12 and Figure j.I:I. This is a
reason that we choose MLE (Maximum Likclihood Est.imate) lIIet.hod to inVl'st.igat."
the actuator models. The relevant comparison curVe is shown in Figure j.I-1.


Comparison Botwoen the Actual and Simulalod Rola1iona\ Angles 01 Swing Molor
150

-100

-15O:------,=::----::':':-----,=::----::!::---:::-::c:---:-:!
o 2000 4000 6000 eooo 10000 12000
sampi. Points (O.OlS/polnt)

Figure ï.14: Comparison Betwccn the Actual and Simulated Rotational Angles of
Swing Motor

• 68
• 1
- = :1.60fi\ (i .:19)

(i.·IO)

1.-.,
-"- = .5:1.0812 (i.·l1)
Rn"

~ = 4.0i·1:1 (i..12)
'1
_1_ = 11.294i (i..1:3)
R't'TI

From (i.:19). we get 'Z = 0.2ii:3. From (i..12), we get '1 = 0.2454. From (i..1:3). we

1
Rn = 0.2-1.5'1 X 11.294i = 0.:36 (i..14 )

Sllhst.it.lltillg (i.·1·1) into (i..1I), we get

• BCCéltlSe T~ = i, wc
k z = 5:3.0812

have

b --
X

!... --
72
0.:36 X 0.24.54 = 4.699i

241.34 - '-03')
')--'3 - ~I
0 ._11· . -
(i..15 )

(iA6)

Sllbstitl1ting (i..16) into (iAO), we get

k, = 0.:3142 X bTz = 0.3142 X 8iO.32 x 0.2ii3 = i5.829 (iAi)

The frcql1ency response plot, i.e. Bode Plot, in discretc-time, is shawn in Figure i.15.
The freql1ency response plot, i.e. Bode Plot, in continuous"time, is shawn in Fig-
ure i.16. There are no rcsonancc (at 2 rad/sec) in these frequency response plots.
The relevant comparison curve is shawn in Figure i.1i.

Note that the above physical parameters derived from the identified mode! are
not neccssarily correct, sincc input-output matching can occur even if the modcls are
different.

• 69
• 10'

'0"
l
1
AMPUTUDE PLOT, Input
r-----~-------- ...
If 1 oulpul
m · ..
If 1

. J';:------~~~~~ -..- .
10° la'
Ircqucncy (rndl~)
PHASE PLOT. input # 1 output" 1
,o~ 10'

l00r-------~------~---~

-200'-::------~;__-----....,.-------'
1~ 10' ,~ 1~
Il'CQuoncy (rad/soc)

Figure 7.15: Frequency Rcsponse Plot or Hoist iVlolor in Discret.(~Till\(,

• 10'
AMPLITUDE PLOT. inpol" 1 Oulput " 1

10. ,°'-;;- "'-;- ":;- ":;-_"""_ _.-.1


1~ la' li 1~ 1~
froquoncy (md/soc)
PHASE PLOT. input" 1 output" ,
50,---,---.,.-----.,.-----...----........,
o
~
{
-50
_100~--'""""'-;---=:::·~-===::::::;::;==--J
10° la' 10 10:1 10"
froquoncy (rndfsoc)

Figure 7.16: Frequency Response Plot or the Hoist Motor in Continuous-Tirne

• 70

Comp~tl:;on Bclwccn tho Actual ilnd Simulatcd Aopo PO~lllon 01 Helsl Mo:or

l
2

l
~ 0
~

li
8.-2
o
a:
-4

-6

-80.'----::2000=---::4ooo=----::6000=---::60~OO:---~10000=:---1=2000
samplo Points (0.01 slpoint)

Figure 7.J 7: Comparison Between the Actual and Simulated Rope Positions of Hoist
Motor

• (:l) Crowd Motor:

2.. = 11.9917
'2
(7.4S)

~
- -_.)_.1~14'0 (7.49)
b'2
ko
D· = 30.4554 (7.50)
'La'l

2.. = 40.6672 (7.51)


'1

Dl
'La'l
= 30.9475 -..
( 1.0_))

From (7.48), wc get '2 = 0.OS3. From (7.51), we get '1 = 0.025. From (7.52), we get
1
D_ -1')9?5 (7.53)
'La - 30.9475 x 0.025 - ._-
Substituting (7.53) into (7.50), we get

k2 = 30.4554 x 1.2925 x 0.025 = 0.984 (7.54)

• 71
• 10
c

10·:;-
AMPLITUDE PLOT. moul" 1 oulpul" l

10~

10"
10' 10' 10:;'
IrCQuoncy (r.ldJsoc)
PHASE PLOT, input # 1 oulput # 1
100

0
0
~
0
~
-100

-200
10' 10' 10" 10'
lroqucncy (rad/soc)

Figure i.18: Frequency Response Plot of Crowd Motor in Discrctl~Tillll'

• AMPLITUDE PLOT. input" output #,


10' r------r-----...,.------~----..,

10·10 ":- .....: - ""':" '"::- ...J


,~ 10' ,~ 1~ 1~
froquoncy (radlsec)
PHASE PlOT. Input. 1 OUlpUt • 1
100,-----.-----,-----,-----,

_100~-----'-:----_'_:"-=:::==::;:;:=:--.....J
1~ 1~ 1~ 1~ ,~
froquoncy (radlsec)

Figure i.19: Frequency Response Plot of Crowd Motor in Continuous-Timc

• i2
• -
B"(";:tll;"f' ';"'., ::::: _,1. \\"t~ hét\'('
,
1 :!(j
/, = - = - - = :11:1')'i
" D.m":l .-
SlIbstitlltill:,!; (ï.'i'i) illt.o (ï.'I!)). we get

l"~, = :!.ïl·l'i X /", = 2.ï1·15 X :~1:~.2'i X O.OS:~ = ïO ..~ïï (ï .'i6)

'l'Ill' freqU('llcy respOllse plot. i.e. Bode 1'101. in discrete-time, is shawn in Figure ï.18.
TIll' ;',equellcy respOllse plot.. i.e. Bode Plot, in continuous-time. is shawn in Fig-
ure ï.I!J. TI",re arc llot noises (at 2 rad/sec) in these fr<'quency response plots. The
rd(~\'atlt. compétrisOIl cur"c is shown in Figure ;.20.

III Figllres 7.14, ï.lï, and ï.20, we can see some large spikcs, which arc duc
t.o shot-type noise, probably due ta large contactor transient on the shovel. These
spikes do not represent the actual characteristics of thcse DC motors. In Chapter 9,
we will sec some spikes also exist in current and voltage signaIs. These spikes are
also duc ta the same shot-type noise. In Figures ï.14, ï.lï, and ï.20, we can sec


good input.-output match, bl,t does not neccssarily indicate correct physical mode!
paraJnctcrs.

CompariSOn BoIWOOll tho Actual and S1mul4tod Arm Extonslon 01 Crowd Molor
'.5

0.5

o
g-o.5
L,
;li
~-,.5

-2.5

-3.50 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
SImple _ , . (0.0,_)

Figure ï.20: Comparison Between the Actual and Simulated Arm Extensions of
Crowd Motor

• ï3

7.4 Parametric Comparison

The De motor paranll'I('rs or thl' analytiea! m",kl \\'('n' ohlaill"" hy II", 1'.\:11 ""111

pany's data Curycs and our cakulatiotl. The De tllntor paralllt'!c!"s (lI' tIlt' t'llIpiricOIl
tll0dcl arc obtained by the field 1.<'st. and Systl'1l1 idt'ntilicat.ioll. Theil' cUll\parisÜlIs ilrt'

shown in Table i.1. In the table. Wl' choosc cmpirieal illl'rl ias are l''1I1;Li to analy! ieal
incrtias. Gccausc the mcchattical conlponcnts aft' reliablc. Fron1 tht'st' t.allIt·s. otlt' cali
find that sorne paramders have similar values. while soml' other paramdl'rs le.~. 1.'1)

have bigger differences. As being indicated in section i.2. although our est.imat.ioll
still converges ta a mode! and input-output. matching cali occur. t.he paraml'l.t·rs or
model can be inaccurate. For el'amp!e, in Table i.1. the swing mot.or u.. = 2.(;:.!·12
is unreasonable high, which will results in low armat.ure currcnt. The rea..' oll or t.his
problem is because wc had not suitable input-out.put data which cali pro<!lIcc t.r:u·
system parameters.. But for our system simulation, the mode!s are good ellough.
.-
2
SwingParameters R.(f!) /(kg.m b kl(N.m/A) k z(v.scc.) Tdscc.) T,( sa.)


)

AnalyticalMode! 0041 24.22 226.:31 2.:38:3 ._,


0')- 0.02·1 O.IUi

Empirica[j\;/ode! 2.6242 24.22 :346 106.ï:3 2.725 O.n:11 !J O.Oi

HoistParameters R.(f!) /(kg.m 2 ) b kl(N.m/A) k 2 (lucc.) Tl (.,cc.) T,( .,cc.)

AnalyticalJ1v/ode! 0.28 241.34 363.1i 6.10:3 :3.411 0.0:16 O.<;i

EmpiricalMode! 0.36 241.34 8ïO.32 ï'::;.829 4.6!J9ï 0.211511 O.2ïï:3

CrowdParameters R.(f!) /(kg.m 2 ) b kl(N.m/A) k 2(v.scc.) TI(SCC.) T2("CC.)

AnalyticalMode! 0.32 26 33ï.12 5.ï26 0.68 0.031 o.on


EmpiricalMode! 1.2925 26 313.25 ïO.5n 0.984 0.025 0.08:3

Table ï.l: Parametric Comparison of Swing, Hoist, and Crowd Motors

• ï4
• Chapter 8

Nonparametric System Identification of Digging


Process

In order to completely understand the shovcl's work and real time monitoring/controlling
of the clectric mining shovcl as weil, the nonparametr:c system identification of the
digging process is imperative. The work consists of two parts. First, nonparametric
identification of the rigid body load; second, nonparametric identification of the both
the rigid body Ioad and the digging process. Actually, the nonparametric identifica-
tion in our research is mainly to identify the armature currents and the positions of


the OC motors.

8.1 Empirical Transfer-Function Estimation (ETFE)

System has inputs u(t) and outputs y(t), t = 1,2, ..., N. One can define [46]:
1 N .
UN(W) = r.; L u(t)e-lI'" (8.1)
vN '=1
1 N .
YN(w) = r.; Ly(t)e-"'" (8.2)
vN '=1
whcrc W = 2rrk/N,k = 1, ...,N. A transfer function based on data over tr.e interva!
1 ::; t ::; N is:
G(
Ne
iw) = YN(W)
UN(W) (8.3)

where G represents the estimation function.


Here, we mainly consider the digging process. When the shovel is digging ore,

• the swing motor is not needed to contribute to the digging work. Therefore, we

ï5
• only stucly armat"r" cnrr"nt for th" hoist and cr<>,,"d molor. l;sing th.. ETFE. \\"('
obtain transfcr function Cttf\"('S ba......cd on input. arrnat.Ufl' \"l)lta~l' and llllt.pnt annat Ufe

current. The transfer function of hoist motor armalnn' an' sho,,"11 ill Fignn' :<.1 alld
the transfcr function of crowd Illotor arnlaturc are shown in Figure ~.~.

8.2 Spectral Analysis Estimation (SAE)

Spectral Analysis Estimation has a smoother estimating curye than tha\. of Empirical
Transfer-Function Estimation. The transfer function is ['16]:

(SA)

The noise spectrum is:

(S.5)

where W.., is weighting function. This method is different from the first method

• because it includes not only the frequency response of the transfer function bnt also
the noise spectrum analysis. The latter part, however, is the important point of
our analysis in the chapter. The transfer function curves are also based on inpnt
armature voltage and output armature current. The transfer function of thc hoist
motor armature is Ehown in Figure 8.3. The noise spectrum of thc hoist motor
armature CUITent is shown in Figure 8.4. The transfer function of the crowcl motor
armature is shown in Figure 8.5. The noise spectrum of the crowd motor armaturc
CUITent is shown in Figure 8.6.

8.3 Lag Window Choice

As the Spectral Analysis Estimation includes the noise spectrum analysis, wc foc us
our research on this method. In spectral analysis, the weighting function W..,(€) in
equations (8.4) and (8.5) is called the frequency window. In our rescarch, wc chose

• the "Hamming" window, duc to av-d.Îlablc MATLAB function.

76
• 10
,
!.

I~~\
AMPLITUDE PLOT. input" , oulput " 1
, ,

1
10-
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 OA 0.45 0.5
Iroquoncy (H:)
PHASE PLOT. input. 1 output" ,
200

100

li:
~ 0
'"
0-

-100

-200
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 OA Q.45 0.5
troquonc:y (Hz)

Figure S.l: Transfer Functions of Hoist Motor Armature (ETFE)

• AMPLITUDE PLOT, Input. 1 Oulpul • 1


10' .--...,....---,.--.,....-....,.--~-..,.......;--,.--.,....-~----,

10'

10~1
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Q.45 0.5
troquoncy (Hz)
PHASE PLOT,lnpul • 1 Oulpul • 1
150

100

~
~
50
{ 0

-50

-100
0 0.05 0,1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
troquoncy (Hz)

Figure S.2: Transfer Functions of Crowd Motor Armature (ETFE)

• 77
• Translor Function 01 HOISl Molor Arm,lluftJ (M .. 1S0) wilh 01991n9
10' r--~-~--'--~--'-~-'--""':'--,::::"''::''''---,

1O'L--=----:':-~:_-::_:_-__:':-___::_-::_:_-__'::-~:_____:
o 5 10 15 ~ ~ ~ ~ ~ 45 W
Iroquoncy (Hz)

Figure S.3: Transfer Function of Hoist Motor Armature (SAE)

• Noise Spoctrum of Hoist Melor Armaturo Curront (M.1 SOl wilh Oigging
10' .--~---.---r--~---.---.---'-~---.--'-'-.---,

10'

10'

10"

10"L_~-"",:"---:,:--:,::-.....",:--:~--::-:----:,:--:,::---!.
o 5 10 15 20 25 30 35 40 45 50
lrequency (Hz)

Figure 8.4: Noise Spectrum of Hoist Motor Armature Current (SAE)

• 78
• Tran!;lot FuncliOn 01
10' r--~--------------"--:"'-----,
C~owd Motor Arm..11urc (M.1S0l wllh Digging

10"7---;---;,;;:--::---:::---,:;:--::;;---,:::---:;;--;;;--;,
o 5 10 15 20 25 30 35 40 45 50
lroquency (Hz)

Figure 8.5: Transfer Function of Crowd Motor Armature (SAE)

• Noise Speetrum ot Crowd Molor Armature Current (M.150) w1th Oigging


10' .--~-...,.--r--~-...,.--r--~-...,.--r--.,

10'

10'

10..7---:---::---;';;---::;---::----,:=---::---;';;--;;;--;
o 5 10 15 20 Z 30 35 ~ ~ 00
'requoncy (Hz)

Figure 8.6: Noise Spectrum of Crowd Motor Armature CUITent (SAE)

• i9
• r t~\\~(':)d': ~

;\l(;):::::~
_:\

'"Y-
.\1(:) (~.ti \

(~.7)

The parameters "1 and Mb) arc used ta evaluate the \\'idth of th,' \\'indll\\'. :\')
decrease (the frequency \\'indo\\' gcts \\'ider). the number M()) innea."". 'l'hl' \\'ider
",indow will include frequencics which may devi,tte far from the ct'ntral freqll<'ncy ~·u.

i.e. bring in a large bias, but lead ta a smail variance of Gu(e''''''). ln order tll 1",\.I,('r
proccss the estimates, wc must trade 01'.' between t.he variance and the bia..<.

In the fol1owing figures (from Figure S.7 t.o Figure 8.12). wc choo't, di 1f<'l"en 1.
M("y), and get ('ifferent noise spectrum estimates. ln our ca.<e, wc chose Al = 150.
which sccms suitable.

Noiso Spectrum 01 Holst Molor Armaluro Curront (M-l00) wilh Digging


10' r--~-'-~-~-~-..,.--~-...,.--'-...,-""":':""":--,


10"

10"~-O--~-"':---:':---'::---'::---'::---':--~_-.J
o 5 10 15 ~ ~ ~ ~ ~ ~ ~
lrequency (Hz)

Figure 8.7: Noise Spectrum of Hoist Motor Armature Current (M:IOO)

• 80
• \0'

10·1
ri
NOI~;U Spuctrum 01 HUI!>l Motor Arm;lluro CUfTlmt (M.:: 150) wlth Digging
---~------~-~----~..:..:..,..:....-~

\0' 1

1
2" ,
.., 10

10·)L--~_-':_-~_~_-':,.-_~_~_-':_ _~_...J
o 5 10 15 20 25 30 35 40 45 50
lroqucncy (Hz)

Figure 8.8: Noise Speclrum of Hoist Motor Armature Current (M=150)

• '0'
Noiso Spoctrum 01 Hoist Motor Armnturo Curront (M_200) with Oigging
.--~-~--,.....-~-~--,.........;.~-~-.:;,:,.,.:...-..,

'0"

,o"~-~-~--::--::,.......--::::---;=---==---:::--::,.......--;l.
o 5 10 '5 20 25 30 35 40 45 50
troque.cy (Hz)

Figure 8.9: Noise Spcctrum of Hoist Motor Armature Current (M=200)

• SI
• 10' l-
i
Noiso Speclrum 01 Crowd Mator Arm."\lure Cunonl (M--100l
10" ri--,..-~,-~--------=,.::-:-~=-~--,

10'

10'

10.,;---;:--:.;-----:;:---;;:----=--=--=-~-__::_-~
o 5 10 1S 20 25 JO 35 40 45 50
lroquoncy (Hz)

Figure S.10: Noise Spectrum of Crowd Motor Armature Current (l'vI = 100)

• Noiso Spodrum 01 Crowd Molor Am'I3luro Curront (M.1SO)


........ ........::,.:............,..;........~-_,
10' r-~-.:.;......:....-;......:.....:..,.....:.-.:- ~

10'

10'

10'

10..L -.......--'--~--'---''--~-~-~-___:" _ _:!.


5 ° 10 15 ~ ~ ~ $ ~ q
lrequoney (Hz)
~

Figure 8.11: Noise Spectrum of Crowd Motor Armature Currcnt (M=150)

• 82
• 10~
,
Noiso SpcC1rum 01 Crowd Molor Armalure Currcnl (M=200)
10~ !-~-~-~, -_-~-~_~-:-~-:-~--,
f j
l
j
10~ .
-)

'O"~---:'-----'.:---:':,----:'::-"""'--'o"-~-~-""""'_...J
o 5 10 15 20 25 30 35 40 45 50
Ircqucncy (Hz)

Figure 8.12: Noise Spectrum of Crowd Motor Armature Current (M=200)

• 8.4 Noise Spectrum Comparisons of Armature Currents


with and without Digging

After we decide the lag window and its M parameter value equals 150, we begin to
analyse the noise spectra of the armature currents of the DC motors. At first, let us
sec why we chose the armature currents as our objective of the nonparametric model
estimation. We recall equ...tions (5.2) and (5.3) as follows:

(8.8)

(8.9)

From the two equations, we have:

(8.10)

• 83
• Under the conditions of certain ~. and ':.:. t hl' chan~e of i" rd\ccts t h"
Thcrcforc. wc conccntrate the tlonparanll'tric ident.i1kat.iol1 in l'st it11<\1.i Il!; l'ach
arnlaturc enrrent.

FraIn aH annaturc eurrent Spt'ctra. wC' <'xt.ract 1.\\"0 t.ypiral parts:


chan~l' ,,1' 1:1•
ll11lh)f ':-;

1l0iSl' Spt'ct 1'11111

with digging proce," and noise spectrtllll withont dig;gin~ pro""" fM hois! ,uuln,)\\'d
motor cases. The relevant results arc shown in Figures 8.1:1. 8.\·1. ~U5. ami 8.\l;.
From these figures. wc can sec: 10\\'er frequency parts ha\'l' a \l\~hcr ~ain aud higher
frequency parts have a smaller gain: there are larger gaius bl'\.\\'l'en :1011" and :1511"
for both motor cases and this indicates the shovel has vibration of non-rigi,l hody in
this frequency range.

Comparing the Figures. wc also find that the noise wit.h diggin~ arc hU'ger t.han
those without digging. This conclusion is in line \Vith the adual circulllstanœ. becau",-'
during the digging process, there are two types of noise for each aduator, i.e. shove!
body noise and digging noise. The differences in the noise spedm just. reflect. t.he
11

• difference between the whole noise spectrum and the rigid body noise spedrum .

• 84
• Noiso Spoctl'1Jm 01 Hoisl Molor Arm.lluro Curronl wllh Diggmg
10' I-~~-~-~-----------~-'

10~
1

10'

10"

°
10"~--=---'::---'::---:',..----:'o-----::':--":---'::----:':,..----:.
5 10 '5 ~ ~ ~ $
Iroquoncy (Hz)
~ g ~

Figure 8.1:3: Noise Spectrum of Hoist Motor Armature Current with Diggillg

• Noise Spcc1rum 01 Hois1 Melor Armaturo Currenl without Digging


10' ,---.--~--..-- ......-~,--_-_-~':"::"'';:'''--,

10'


Figure 8.14: Noise Spectrum of Hoist Motor Armature Current without Digging

85
• 10'

10' t
:
Noiso Spcctrum 01 Crowd Malor Arm..1:urC Cuneol wilh Dlf19lng
r-~--'--~-~-~-~-~-~-"--~'~--l

10'

10'

'o-:L-_-'-_~_·~_~_-' _ _'-_~_~_-':_-='
o 5 10 15 ~ ~ ~ ~ ~ 45 ~
frcqucncy (Hz)

Figure S.1.5: Noise Spectrum of Crowd Motor Armature Currellt. wit.h Diggillg

• '0'
Noise SpDdtUm 01 Crowd Molor Armature Currant without Dïgging
.--,..--..;..----r-~-~~-~-..,..._-..,...=..::,.-_,

'0'

,o.,L--~-_'_:"-_'_::-__:':,.....__=':__,'_:_-~-_':_-__'_:,.....__=
o 5 10 1S ~ ~ ~ $ ~ 45 ~
!_oncy (Hz)

Figure S.16: Noise Spectrum of Crowd Motor Armature Current without Digging

• S6
• 8.5 Noise Spectrum Comparisons of Armature Currents
with Easy and Hard Digging

The '';L'Y alld hard diggillgs cali also have dilferellt elfeet on the armature eurrents.
dlle tü differclIt interactional forces. The relevant results arc shawn in Figures 8.1i,
S.l S, 8.1!J, and 8.20.

COlllparing the Figures, wc also find that the noise with hard digging is larger
thall those with e""y digging in low-frequeney parts, but smaller in high-frequeney
parts. We believe that during the hard digging process, due ta harder ore, interaction
has slower proeess - this brings in larger low-frequeney noise (impact) and smaller
high-frequeney noise into the armature eurrents.

8.6 Noise Spectrum Comparisons of Positions with and with-


out Digging

• ln addtion ta armature eurrent noise spectra, wc ean estimate the position noise
spectrum as welt. lt fully rellects the aetual noise e[ect of the motor. Figures 8.21
and 8.22 show the noise spectra of the hoist motor rope position. Figures 8.23
and 8.24 show the noise speetra of thn erowd motor arm extension, when wc set
M = 150.

The noise speetra of the positions have similar features ta the noise speetra of
the armature eurrents, except that the gains between 25Hz ta 30Hz and 40Hz ta 50Hz
have larger values. That means the shove! has also vibration of non-rigid body in the
two frequeney ranges. From this phenomenon, we can recognize that the e[ects of
non-rigid body truly exsit on the e!ectric mining shove!.

• 8i
• 10'

10'
Noise Spoetrum 01 HOlst Melor Arm."\luro Curront Wllh EOlSY Ol9gin9

-\
10:
1
10'

10'

10·'

10"

10"
0 5 10 15 20 25 30 3S 40 45 50
froquoncy (Hz)

Figure S.l i: Noise Spectrum of Hoist Motor Armature Current with Easy Digging

• 10'

10'
Noise Spedrom 01 Halst Mater AAnaturo Curront with Hard Otgglng

10'

10'

10'

10.'

10-0

10"
0 5 10 15 20 2S 30 3S 40 45 50
lrequency (HZ)

Figure 8.18: Noise Spectrum of Hoist Motor Armature Current with Hard Digging

• 88

NO'50 Spcetrum 01 Crowd Meler Armatur~ Currcnt Wlth Ea:.y Digging
10'
1
10'
1 ~
'0'

'0'

'0'

10·'

10·"

10.3

'O~
0 5 '0 '5 20 25 30 3S 40 45 50
froquoncy (Hz)

Figure 8.19: Noise Spectrum of Crowd Motor lumature Current with Easy Digging

• 10'

10'
Noiso Spoetrum 01 Crowd Moter Atmntur'O Currant with Hard Oigging

'0'

10'

'0'

10- 1

10"

10"

10~
0 5 10 15 20 25 30 3S 40 4S 50
lroquency (Hz)

Figure 8.20: Noise Spectrum of Crowd Motor Armature Current with Hard Digging

• 89
• 10~

10'
Er,
Noiso Spectrum 01 Hoist Molor Posil ,.:ln Wllh OigC";; I~j
-~-~--,-~-~-~-~-~-,-.-.-;--,~!

10'

10'

10-:;---:----:'::---',--:::_~~-":_-"_-..,.,.---'::__:'
o 5 10 1S ~ ~ ~ ~ ~ 4S ~
Irnquoncy (Hz)

Figure 8.21: Noise Spectrum of Hoist Motor Rope Position with Digging

• Noise Spectrum 01 Heist Motor Posilion wilhout Digging


10' ,--.--...,...--.--~--~-..--~-'-"-,--~-
...

10'

10'

10'

10'

5 10 '5 20 25 30 35 40 45 50
'requoncy(Hz)

Figure 8.22: Noise Spectrum of Hoist Motor Rope Position without Digging

• 90
• 10'

10'
!----.---
,
1

i
i
NOI!'.f} Spcctrum 01 Crowd Mot:,..r PoslliOn Wllh Olgging

10'

10'

10->L_~_",-_~-~-~-~-~-~--,::--...,J
a 5 10 15 ~ ~ ~ $ ~ ~ ~
lroquoncy (Hz)

Figure 8.2:l: Noise Spectrum of Crowd Motor Arm Extension with Digging

• Noise Spoctrum 01 Crowd Melor Position wilhout Digglng


'o' ,....--,-....:...:;:::..::.:.:,..::.-.:...;.:.::....::.-;:.:...:...:.:,:.:.....:...:;:::...::...::;....:.......,.-----.

10'

10'

10'

10..'L--'----'---'"-~~-'--~_--'---'-
_ __'_ _
o 5 10 15 ~ ~ ~ $ ~ ~ ~
lroquoncy (Hz)

Figure 8.24: Noise Spectrum of Crowd Motor Arm Extension without Digging

• 91
• Chapter 9

Simulating Methodology and Simulator

For the purposes of building a softwar" simulator. ail Ut<' mod"is d"ri\'('d iu th" l'n'\'i-
ous chapter must be integrated in a coherent manner which r"l\,·d.s tlt<' h"h""i"lIr "f
the actual system, while remaining computat.ionally efficient.. Sll\IULINK. a silllula-
tion toolbox of MATLAB, is used in our syst.em simulat.ion. Th,' f\"nl'ral simulat.i"n
system structure is shown in Figure 9.!.

9.1 Overall Machine/Process Model

• The overall machinefproccss modcl includcs the actuat.or dynamics, rigid hody dy-
namics, and the interaction betwccn bucket and muckl'ile.

In terms of the thrce actuators, one can obtain crowd, hoisl. and swing lIIol.or
dynamics, respectively. Using equations (5.10), (5.11), and (5.12), wc have

(!l.1 )

(9.2)

(!l.:l)

Actuator ~ Multibody
Dynamics Dynamics

• Figure 9.1: Simulation System Structure

92
Consid"rin!\ th" threc cliffcr"nt parts, one can obtain the :natrix representation

• fortII ;:L~ follows:


XI

x~

;r.:~
[)

[)

[)
_.!.

-~
~
[)

-.b...
111'T:z
_.!.
[)

[)

[)
[)

[)

[)
[)

[)

[)
[)

[)

[)
[)

[)

[)
[)

[)

[)
Xl

X2

X3
Il.. ,,! Tl

X" [) [) [) [) [) [) [) [) x.,
d _.!. k,
dt
x'
' .. = [) [) [) [)
T, ~
[) [) [) X5

XG [) [) [) [) _...b- -.!. [) [) [) X6
llo'2T'J T:l
.,
x- [) [) [) [) [) [) [) 1 [) I7

[) [) [) [) [) [) [) _.!. ~
k.
IS
XM TG b31'G

x~ [) [) [) [) [) [) [) _...10- -.!. I9
HQ3"~ ",

[) [) [) [) [) [)

[) [) [) ...!:l... [) [)
b,~
1
Iloi Tl
[) [) 0 0 0
0 [) 0 0 0 0
+
C'::l - 0
C"l
Td2 (9.4)
0 0 0 0 ..!:Z-


b, T 4
t
0 R.,T:l
0 0 0 0 Td3
0 0 0 0 0 0
0 0 0 0 0 ..!:Z-
b:rTG

0 0 R.,,,,t
0 0 0
The gearbox rates, rI, r2, and r3 are bro'lght i!ito the system simul",tion. This is
refiected in position and velocity outputs as well as disturbance inputs.

9.2 Continuous-Time Controller

The rotational velocities of all DC motors are controlled by analog electronic control
circuits on shovel. In order to the purpose of simulation, one can design the controllers
in terms of the electrical parameters of the three motors.

1. Analytical Model Controller


Recalling the pararneters in Section 5.2, we have:

93

Swing :-'lotor:

l?" = 0.·11 P.. 1 = :?\.:?:? 1."!1.11l~. b = :?:?t;.:; 1

T, = 0.02·1 .'<:C.. T~ = 0.1 Oi -'<"C.

Hoist Motor:

kt = 6.lOa iV.n> / A. k~ = :l..14 " .scc.

Tl = 0.Oa6 sec., T~ = 0.6; scc. ( ~l.li)

Crowd Motor:

Ra = 0.a2 fi, 1 = 26 kg.m~. b = :1:1;.12

• k l = 5.;26 N.m/A,

Tl = O.Oal sec.,
k~

T2
= 0.68 v.see.

= O.Oi; sec. (9.7)

In order to design the controller, we change (5.10) and (5.11) into:


. I kl r T.
Xl = --XI
T2
+ -bT2
X2 - -
bT2
d (!I.S)

(g.!))

y =Xl (9.10)

The matrices are obtained by inspection of the above equations as:

A= ( -~ ~) ( 0 )
_...s- :~ , B = _1_ ' BI =
(_2:...)
;b (9.11 )
RaT, 'TI R..'T1


C=(I 0) (9.12)

94

(1) Swing Motor Controller and Disturbanee Compensation

SlJ1lstilliling t Ill' p;trillllt'l.l'rs frotll (,"llatioll (!L;j) illta (!J.ll). \\'(' han'

1) '1"
,'1 =
- .. ' .)
().{)!J~ )
.U= ( 0 )
(
-:!7"J,1 -'II.fi7 10 1.6:l

I.t',.

X3 =y- r = Xl - r (9.15)

The .·i and Ù mat.riee, arc

-9.:35 0.098

• A=
(
-2; ..14

The cont.rollabilit.y mat.rix is given by


-41.67
o
(9.16)

9.9597 -508.1456)
-4234.9221 1i6195.91 (9.1i)
o 9.959i

The dcterminant of Q is nonzero. Its rank thp.reforc is 3, and the pair (...i, B) is
cont.roIlablc. The eharact.eristie equation of the closed-loop system is given by

s + 9.35 -0.098 o
IsI - ..\ + Bkl = 2i.44 + 101.6:3kll s + 41.6i + lO1.6:3k 12 101.6:3k13
-1 o s

= + (51.6297 + 101.63kds + (:392.3036 + 950.2405k12 + 9.9594k


S3
2
ll )s + 9.959ik13

• 95
(9.18)

l'hl' de~irt'd charactl'rist ie ('quat iün i:,

."
:3 1
-;
~-
~j .:J .... -
- .,
î
, -'1- -
;),)I·1.~) .... i
'1'"_·l~)h~l = l)

:-'latching cqllatiotl' (9.18) and (9.19). we ohtain

In the first part of thc eqllation (9"1). th,' dis! nrhalll'<' conqH'lIs"t iOIl h"s th,' r"lI"win",
form
1
-9.:>5
-27.44
In the condition of steaely-state. wc have il = O..iz = Q,and i" = O. TIll'refon·.

-9.:>5x; + 0.098x; - =Q
0.0'11:lr l 'r:il
-27.44x; - 41.69x; + 101.6:kj = 0 _-
( ').. "")

Solving the above equations, we have

• v "t =.).
"9 40-- "+ 0• 1-')9
IIX 2

By the state feedback law, .6.v = - K .6.x, we have


1- Tl 'r"
.. ,ll

VI = 39.4077x; + 0.1729rl Td1 - 0.3529x3 + (466..58 + 12547.466


, )(x; - xz) (!1.2'1)

(2) Hoist Motor Controller and Disturbance Compensation

The  and ÎJ matrices are


0.Q25
-27.78 'r)
(Il...... :)

o
The controllability matrix is given by

(9~.2
2.48 -72..5896 )
Q=(B ÂB Â'B)= -27.55.776 7.5709.108 (9.26)

• 96
o 2.48

li,.. dd"r'llil""1 "f Cl is ",,",,'ro, lis r,,"k IIll'rt'fort, is :\. ,,"d th" p,m (:1. 13) lS
l·f)tlt.rollttl)lt" TIlt' cil;lractf'risl je ('qllatioll or tlw closed·loop syst.em is giVPIl by

,,+ lA!) -0.025 a

-1 a .s

= ,
,:1 -l. (')l) .,-
tT nI)
, J "k )
· · _ 1 5 · ,2 + (11) 9')1 T' 1'1-
'I.ul,.; ,.) 1~k l·l~T_.·u
~Osk 15T-.~o )' , .) 1~k 16 (9.2ï)

:'\'Iat.('hillg ('quat.iolls (9.2ï) and (!1.19). we obtain

k", = 2112.021:3. k,s = 0.58ï. k 16 = .50:390.ï26 (9.28)

ln t.he second part of the eqnation (9.4). the disturbancc compensat.ion has the fol-
lowi t1~ [onn

i.,) (0 1
( i" = 0 -1..19

• .,'" 0 -:)41.2ï

ln t.he condit.ion of st.eady-state. we have X.I

-1.49x[; + 0.02.5x~ - 0.0041r]d2 = 0


= O.xs = O.and X6 = O. Therefore.

-:341.2ïx[; - 2ï. ï8x~ + 99.2\1;- = 0 (9.:30)

Solving t.he above equations. we have

V z = 20.1921x[; + 0.046r2Td2 (9.:31)

By the state feedback law. ~V = -/{~x. wc have


"2 = 20.1921x[; + 0.046r2Td2 - 0.5Sïx6 + (2112.0213 + 50390.i26 )(x[; - xs) (9.32)
s

(3) Crowd Motor Controller and Disturbance Compensation

The ." and Î3 matrices are

• 9i
• ..l=
o
0)o
o
./1= (0)Illll.~1

The controllabilit.y mat.rix is gin'n hy


- 100:l.:ili:;.\ )
IO:I:l!I:U pUI)
o :!:.!.l 7~~

The determinant of Q is nonzero. It.s rank t.herefore is :1. :11I<1 t.lu· l'ait" (:1. Ii) is
controllable. The charact.eristic equation of t.he c1ose<l-loop syst."1II is gi\"'n hy

s + 12.99 -0.22 o

-1 o

• = S3 + (45.25 + 100.8Ik1s )s2 + (4:3·1.1:384 + 1:309.5219k ls + 22.1 782k 1d" + 22.1 7S2/"I"

Matching equations (9.:35) and (9.19), wc obtain

kt. = 198.0116, k lS = 004191, kw = .56:34.7675


Ul.:l!i )

(!l.:W)

In the third part of the equation (904), the disturbancc compensation has t.he followinl';
form

( ::) = (~-1~.99 0:2) (::) +( ~ ) U3 + (-0.~:38r:l) '[:1:1 (9.:17)


X9 0 -68.55 -32.26 X9 100.81 0
In the condition of steady-state, we have X. = O,xs = O,and Xu = O. Therefore,

-12.99x;; + 0.22x; - 0.0:38r3T.i:l = 0

-68.55x;; - 32.26x; + 100.8lu; = 0 (9.38)

Solving the above equation, we have

• U3• = 19.'J--4· • + 0.0·-3


1 .)Xs 'J'J. r3 T-
d3 (9.:39)

98
• 1', = 1!J." 7·1".r~ + 0.0"":;",, 'I:i" -

2. Empirical ModeI Controller


O..-lI!J I.r" + ( HJ8.0 Il G +
"G:l·1. 7G7"
....
)( .r~ - ;1',) UJ.·IO)

!l<'""llillf; 1.11<" p"r"lIIet.cr, in Scct.ion 7.4. wc havc:

Swillg :\-Iot.or:

!l" = 2.G2·12 n. 1 = 2,1.22 kg.m z• b = a'lG

k, = 10G.7a N.m./A, kz = 2.72.5 l'.sec

Tl = 0.0:319 sec., TZ = 0.07 sec. (9.41)

1I0i,t. Mot.or:

R" = o.aG n. 1 = 241.a4 kg.m z, b = 870.a2

• kt = 75.829
Tl
N.m/A,

= 0.24.54 sec.,
k z = 4.6997 l'.scc.

TZ = 0.27n sec. (9,42)

Crowd l'I'Iot.or:

Ra = 1.2925 n, 1 = 26 kg.m z, b = 313.25

kl = 70.577 N.m/~1, kz = 0.984 v.sec.

Tt = 0.025 sec., TZ = 0.083 sec.

The mat.rices are obtained by inspection of the above equations as


_.1.
·1 -
• - (
-~
':>
1..... -:'~
b.
Tl
)
,B 0 )
= ( _1_
RoTt
,BI ;b
= (_...!:..) (9.44)


C=(I 0) (9.45)

99

(1) Swing Motor Controller and Disturbance Compensation

SlIbstitllting the empirical parallll'tl'rs from (lUI) iUtll th"I'quatillu \l1.l1), \\'1' haw

..1= (
-1,1.2.1,12 G.G:l(2)
.U=
(Il) (lUli)
-:12.5·10:3 -:II.:l:I.1~ 11.\)·11

\Vc dcfinc an additional st.at.c variable .l":t a~

X3 = 10' (y - r)di

I.e.

X3 = Y - r = XI - r (!l .. I~)

The r1 and Ê matrices are

• 6.6:362
-:31.:3:3.')8
o
(!).'1!))

The controllability matrix is given by


ï9.2429 -:l612.6ïï5 )
-:3ï4.18 9146.6419 (!l..'iO)
o ï9.2429

The determinant of Q is nonzero. lts rank therefore is :3, and the pair (Â, Ù) is
controllable. The characteristic equation of the c1osed-loop system is given by

s + 14.2542 -6.6:362 o
IsI-Â+Êkl= :32..540:3+11.941k21 s+:31.:3:3.58+11.941k22 11.941k23


-1 o s

100
_ ." (F.. 'i!! + Il.!!l JI.."" j.," + ((j(i:!.(j 1:l') + l'(J.:!O!J·ll.-"" + '!J.:!·1:!91.-"d.' + ,9.2·129/"""
• (9.51)
'l'Ill' d(·sin·d cilar'Lct.('ristic (·qllat.ion is

(9.52)

\1"l.chill!\ (·'!lIal.iolls (!J.'i 1) alld (9.;'2). we obl.aill

(9.5:3)

III 1.1", firsl. parI. of I.h.. e,!lIatioll (9..1) t.hc disturbance compcnsation has the following
forlll

III t.he rondit.ion of st.eady-state, wc havc XI = O,X2 = O,and X3 = O. Thcrcforc,

11• • ')51')
_" _.T • + 6•6"6')' .).,- T-dl-
0._.J,rl - 0


- 2 •) _x 3 -

-a2..540ax; - :H.a:3.58x; + 11.941xî = 0 (9..5.5)

Solving t.hc abovc e,!uations, wc havc

(9.56)

By the st.at.c fccdback law, ~v = -h'~x, we have


1--- OT')
/'1 = 8.:36:31x; + 0.09a7rl Tdl - a.5098x3 + (51.9224 + .')/1 •• /- )(x; - X2) (9..57)
s

(2) Hoist Motol" Contl"ollel" and Distul"bance Compensation

SlIbstitllting thc cmpirical parametcrs (9..12), into the equation (9.44), wc have

-3.6061
.\- 0.3142 ) ,B= ( 0 ) (9.58)


. - ( -53.0812 -4.0743 11.2947

101

The ,·i and B mat rin', arl'
lUII:!
-Hl'·t:; 0)o . = (0)
Il 11.:!91'
o o 0

The cont.rollabilit.y mat.rix i, gi"en hy

-:!ï.:!:ili:! )
-·16.01 ~ -lU;~:;;) (\l.tiO)

o :t:;·I~~

The determinant. of Q is non zero. 11$ rank t.herefor<' i, :1. and t.he pair ( ..i. Ii) i,
controllable. The characteristic equat.ion of t.he c1osed-loop systt'm is gi\'<'n hy

s+ :3.6061 -0.:11·12 o
Isl - .-1 + Bkj = 53.0812 + 11.294ïk2., -' + 4.0ï·1:1 + \1.294ïk2r. 11.2\Hïk2.:

-\ o

• l\'latching equations (9.61) and (9 ..52), we obtain

(!1.1'2)

In the second part of equation (9.4), the disturbance compensation h'L' t.he following
form
1
-3.6061
-53.0812
In the condition of steady-state, we have X., = D,x, = D,and Xl; = O. Therefore,

-:3.6061x;; + 0.3142x~ - 0.02.56r21d2 = 0

-53.0812x;; - 4.0ï4:3x~ + 11.294ïv; = 0 (!J.(j4 )

Solving the above equations, we have


(!Wj)

102
• ", = ~.~'l.r:' + O.02!J·1r·;I;j, - ï.OGï.r,;

(3) Crowd Motor Controller and Disturbance Compensation


+ (1·1IG.9 +
:1521·1.·1·1
-'
)(:r:' - x,,) (9.GG)

SlIhst.it.llt.iug paralllct.ers ill e"nat.ion (9.'1:3) iuto (9.44). wc have

2.ïl'15
-'1O.66ï2 (9.6ï)
o
The coutrollability matrix is given by
84.00ï -442:3.ï2)
-1258.5.5 4862:3.24 (9.68)
o 84.00ï

• The dderminant of Q is nouzero. 1ts rank therefore is :3, and the pair (...1, E) \s
coutrollablc. The characteristic equation of the c1osed-loop system is given by

s + 11.991ï -2.ï145 o
1.'1 - .·i + J3kl = :lO.4554 + :lO.94ï.5kz• s + 40.66ï2 + :30.94ï5kz8 :30.94ï5kz9
-1 o
= .," +(52.6.589+:l0.94ï5k z8 )sz+ (.5ïO.:34 +:m.11:31kz8 +84.01kz• )s+84.01kz9 (9.69)

i\latching eqnations (9.69) and (9..52), we obtain

k z• = 52.2124, k Z8 = 1.1258, k Z9 = 148ï.5491 (9.iO)

ln the third part of the equation (9.4), the disturbance compensation has the following
form

(
::) = (~ -11~991i 2.i~45) (::) + ( ~ ) V3 - (0.0:r 1 ) Td3 (9.il)

• X9 0 -:30.4554 -40.6i X9

103
30.95 0
• -11.!J91 "r~
-:l0.'1:;5·1.r~ -
+ ~,'I·I:;,r,;
·IO,(i(i'~J',;
-l1,l1~r,.r,i"

+ :10,9·1,,,,',;
= 11

= 0 ... -
l I) -")

Solving the above equations. w<' have

( 1)
•• -")
1 .}

By the state feedback law, .:lL· = -I\·.:lx. we have

+ ("') 1'18""'191) ( ," ,) _)


. - 6.,,, . ,,,'1"3 '1"</3- 1• 1')-"
-"9'3,xs"-1-' 00'3"- • •' +
;)• .')\')'1 1
V3- • '.)"X"
, ,ls-,ls (1)
"f'

9.3 Saturation

Due to the motor and power amplifier characteristics, t.he input. current. and 1,\,'n,-r...I.<-d

• torque cannot surpass certain values. Saturation funct.ions on both t.1",se variables
are used to replicate this behavior in the simulation. The limit.s on t.h" input. "lIl'n'nt.
and generated torque effectively limit the maximum velocity and accclerat.ion of 1.\",
motor.

9.4 Filter

The simulated trajectories of three OC motors had similar shapes to the ad.llal t.ra-
jectories, but the armature voltages and currents included a lot of high-frecl'1"lIcy
disturbance. In order to solve these problems, At first, wc used FFT t.echniqlle t.o
check the frequency ranges of the actual voltages and currents. Then, according t.o t.11l'
ranges, we designed the low-pass filters. Finally, we used the filters in ollr simlllat.or
for improving the output curves of the armature voltages and currents.

• 104

9.5 Disturbance

'l'Ill' disl.lld,atlC'('s t.() t.lw act.llalors primarily arise from 1.\\'0 sources: the rigid body
rrc,ss ('ollplitl~ 1)('1.\\'(~(~11 COtIlpolu'nts of 1.11C' shovd it.sdf and the interaction bct.wcen

1III ,"J.:"r. "~,,I 11I1I<Okpil,' durillg the course of digging and loading. For the first source.
IV(' ,""n liS<' forllllll"e ('I.!i:I). (.1.!i6). and (·1.62) 1.0 represent the disturbance. For the
SPco11C1 S()lIrce~ the distllrbance is alrcadv includcd in il ~ ;3_ and ;.1. idcntificd as f51
"",1 J"" in fOrll1l1l"" (6..15) and (6..16).

9.6 Analytical Model Simulation

Ali three parts of the shovel simulator arc integrated in a single simulation file which
is shawn al. its most superficial level in Figure 1 in Appendix. The complexity of the
o"erall software simulator is illustrated by examining the subroutine blacks which arc
shawn in Appendix. Figure 9.2 and Figure 9.:3 show the rotational angles and voltages

• of swing motor. Figure 9.:1. Figure 9..5. and Figure 9.6 show the rope positions.
voltages, and currents of hoist motor. Figure 9.ï. Figure 9.8, and Figure 9.9 show
the arm extensions, voltages. and currents of crwod motor. From these figures. wc
can see the actual and simulated curves of trajectories and voltages have very similar
shapes; the basic shapes of the currents are also similar.

9.7 Empirical Model Simulation of Easy Digging

The empirica! model simulation of easy digging has the same block diagram as the an-
alytical model simulation except that the motor parameters are different. Figure 9.10
and Figure 9.11 show the rotational angles and voltages of swing motor. Figure 9.12,
Figure 9.13. and Figure 9.14 show the rope positions, voltages, and currents of hoist
motor. Figure 9.15. Figure 9.16. and Figure 9.1 ï show the arm extensions, voltages,
and currents of crwod motor. From these figures, we can sec the actua! and simulated

• cun'cs of both trajectories and voltages have very similar shapes; the basic shapes

105

of the C'urrcu1,s are abo silnilar. In actual applicatioll. wc pref('r l(~ Il~(' tht' l'illpiril';d
rnodcl . duc to that il. bcttl"r rl·T)fl·~t.'llb t hl' aet Hal :-;Ît.nat jtltl.

9.8 Empirical Model Simulation of Hard Diggillg

The empirical mode! simulat.ion of hard digging has sam,' hlock diagram as 1 hal uf
easy digging except t.he syst.em ha,< dirrert'nl illpnt. ,Iat.a. Fignre !U~ alld Figm,' !U!l
show the rotational angles and volt.ages of swillg mot.or. Figme !l.:!O . Fign n' !l.:! l, and
Figure 9.22 show the l'ope posit.ions, \·olt.ages alld currenls of hoist. molor. Figlll'" !1.:!:I.
Figure 9.24, and Figure 9.25 show the arm ext.ensiolls. volt.ages, and CIII'1"'IIt.S uf
crowd motor. From these figures, we can see t.he aetual alld simlllat.ed clln'es of I,,'lh
trajectories and voltages have very similar shapes: t.he adllal alld simlllat.ed clln'es
of eurrents have sorne differenees, but the basie shapes are similar.

In general, from ail simulations in abo\'e thrcc sections, we cali find lbal lb,-


trajeetories are weil simulated; the voltages are a little bit worse simlllat.<-d: and t.Il<'
eurrents are worse simulated, beeause the actual trajectories have a liule bigb fn-·
queney noise; the aetual voltages have more high frequeney noise; t.he adnal clIl'rent.s
have the most high frequeney noise (in our simulators doesn't. exist. ill lbis kind of
high frequeney noise) .

• 106

Actual ROlational Angle 01 Swing Molor
~ 150
~ 100-
1

~ 50f
~ O~. . .r
.~ _sol
;g _,OOIL_--c:':'::,...._-~---::"::-:----::'::----:7.~--~
o 2000 4000 6000 8000 , 0000 , 2000

5imulalcd RotationaJ Angle 01 Swing Motor (Analytical Model)

~ 'SO'--I~~~~~-'i.
f'OO
j 50

"c 0
.2
ëii -50
15
lI:
-100 L...._ _:-:'-:::--_--,-,'::-:-_-,:::':-:-,_-:::'::-_---,:-:-:,:--_-:-::'.
o 2000 4000 6000 8000 10000 12000
Samplo Points (O.01s1point)

Figure 9.2: Actual and Simulated Rotational Angles of Swing Motor


Actual Voltige of Swing Melor
1000..---,....,....---,--

-looo:----,:-:'-::;__-~'::-:--__::::':_:--~::--.....,,::-::;__-~
o 2000 4000 6000 8000 10000 12000

Simulaled Voltago ot Swing Motor (Anolytical Model)


'000r---~---_,_---.:---~---_,_--__,

- 500
~
8,
m
'8
>-500

-1ooo:----,::':::;__--::'::-:--__::::':_:,...---:::'::--.....,,::-::;__-~
o 2000 4000 6000 8000 10000 '2000
$ampl. Points (O,O's/point)

Figure 9.3: Actual and Simulated Voltages of Swing Motor

• lOi
• Aetual Ropc POSitIon 01 HOlst Molor

Simulatod Ropo Position 01 Heisl Molor (Analytie:tl Modol)

2000 4000 6000 6000 10000 '2000


S3mplo Points (O.Ols/point)

Figure 9..1: Actual and Simulated Rope Positions of lIois\. Mo\.or

• ~
1000r---~---~---r---~---~--
Adual Votlagc of Heist Motor

f
> -500

-looo:------:::':::----:::::-------:::':::----::=::--"'"""7:::::---_=!.
o 2000 4000 6000 6000 10000 12000

Simulaled Voltage ot Haist M010r (Analyticat MOdeI)


1000

!..
..'"
~ -500

-1000
0 2000 4000 6000 6000 10000 12000
Sample Points (O.Ols/point)

Figure 9.5: Actual and Simulated Voltages of Hoist :Vlotor

• lOS
• AclunI Currcnl 01 HOlst Mo:or
._~--~-----,

o 2000 4000 6000 8000 10000 12000

Simutatod Curro"! 01 Heist Meler (Analytical Modol)

-1000
0:--------::2:f::
000:-----:::. 0 --:6000=:::--8:::':000::-----:-:,00=00:----;;:,2000
00::-
Samplo Points (0.01 slpoint)

Figure 9.6: Act.ual and Simu!ated Currents of Boist Motor

• :[2
.~ 1
Aetual Arm Extonsion of Crowd Meler
3r--~-""'--~--~--"""---'

a
~ 0
E
<-1
2000 '000 6000 8000 10000 12000

Simula1ed Ann Extonsion ot Crowd Moter (Anillytical Modol)


3r--~-""'--~-~----=--...-'-----'

:[ 2

.~ 1
a
Ji 0
E
<-1
2000 '000 6000 8000 10000 12000
sample Points (D.Ols/point)

Figure 9.7: :\ctual and Simulateà Arm Extensions of Crowd ~'1otor

• 109
• o
~
m
o
Aetual Voltago ,,' Crowd Mc:or

'"
> -Soo

-1000
o 2000 4000 6000 8000 10000 12000

Simul310d Vollago 01 Crowd Molor (Analytical Modol)


1000,----~--~---~·---~---~

SOO
~
g, 0
m

'" -Soo
>

-'OooL----::':':_:_--~--~':':_--='=_:_-___:_::_:_::,_-_:_:_'.
o 2000 4000 6000 8000 '0000 12000
Samplo Points (O.Ols/point)

Figure 9.S: Actual and Simlllated Voltage,; of Crowd iVlot.or

• Ac:tual CUlTent 01 Crowd Melor

-400
0:-----:2::000::----:4-=000=----:6::::000::----=80=00::---:'-:00=00::---:'-:::2000

Simulatod Currcnt 01 Crowd Molor (AnalytiCôlI Modol)


6oo.------r---~---r---_,.---..,..--__,

400
g 200
ë
ê 0
<3 -200
-400
0:-----:2::000::----:4-=00:::0,-----::6000=--""'-:80=00::---:'-:00=00::----:'-:::2000
samplo Pain.. (0.0' slpolnl)

Figure 9.9: Actual and Simulated Currents of Crowd Motor

• 110

Aclual Ao:allonal Angle 01 SWing Meler
150 . --,
ô 1
,
0
J
~ '" 100~
!"
~
c
..:
sof 1
" -~f
" c
Q 1
"5
c:: 1
-'00
0 2000 4000 6000 6000 10000 12000

SimulOltCd ROIOllionOl! Angle 01 Swing Melor (EmpiriC31 Model)


'sor---~---~---~---~---~---"
î ~100

E
.Q 50
~

~
'"
c 0
.~ -50
~ -1 00 L-_ _,..,..,,...-_ _"__----,,,..,..----:-!__---,:-:-:::__-:-::!
o 2000 4000 6000 8000 10000 12000
S01mpto Points (D.Ols/point)

:.'igllr<' n.IO: Actnal and Simulated Rotational Angles of Swing Nlotor (Easy Digging)

• -
!
g.
500
AetUOlI VoltOlg" of Swing Motor
'000r---~---__.---~---..,..---~--___,

';;
> -500

-,ooo~-_.,::':::__--==_--::=---:-::::_---::=:__-:-::'.
o 2000 4000 6000 6000 '0000 '2000

SlmulBtod Voltage 01 Swing Moter (Empirical Modal)


'000.-----,-----,----,..----,-----,----,

I 500
!l,
~

';;
> -500

-,ooo~-_.,::':::__--==_--::"..,..----:-!=_---,:-:-:::__-:-::!
o 2000 4000 6000 6000 10000 '2000
Sampi. Points (0.01s1poinl)

Figure 9.11: Actual and Simulated Voltages of Swing Motor (Easy Digging)

• III
• lc
.2
.~

~
AClual Ropc PoSilion 01 Haist Malor

8.
o
Cl:

Simu13tod Rop~ Position 01 Heist Melor (Empiricnl MOdol)

2000 4000 6000 8000 10000 12000


sample Points (O.Ol slpoinl)

Figure 9.12: Actual and Simulated Rope Positions of Hoist Ivlotor (Easy Digging)

• !
AetU31 Voltage 01 Heist Motor
1ooor---~--~-----=~--"""---~-----'

f
>-500

-1ooo'-----'-----'---__'---~---_'_:_--·-'
o 2000 4000 6000 8000 10000 12000

Simulatod Voltago ot Heist Motor (Empiricnl Modol)


10oor-----.----~--=.--r---...;.".-----=~--_,

-1000 '----2000-'-,...---4000-'----6000......---80-'-00---1-00"'00..,...---12000"
0
$amplo Points (O.Ols/point)

Figure 9.13: Actual and Simulated Voltages of Hoist Motor (Easy Digging)

• 112
• AC1ual Currcnl 01 Hoist Molor

-1000
0~----'2:00'::OO'-----4:-:00:':0---'---:6OOO=---:8:::000::----:-:'OO::OO:::--~'2ooo

Simulalod Curran! of Hoist Molor (Empirical Modol)

-'000
0~----'20:-:'::OO'-----4:-:000::-----:6OOO=---:80:::00::---:-:'OO::OO:::----='2ooo
Somplo Points (0.0' sJpoinl)

Figure 9.14: Actual and Simulated Currents or Hoist Motor (Easy Digging)

• Actual Arm Extonsion of Crowd Molor


3r-----.,..---..,.----.------,---....,..---,

Slmulatod Arm Extonslon 01 Crcwd Molor (Empirlcal Model)


3,-----,..---.-----.------.---.......----,
g2
.~ 1
c
;li" 0
E
<-1
2000 400D 6000 8000 1‫סס‬oo 12000
Somplo Points (D.Dls1po1nt)

Figure 9.15: Actual and Simulated Arm Extensions or Crowd Motor (Easy Digging)

• 113
• ~
o
~
'OOO,.---~---_----'-,.---~-

500

o
Aetual VoIlage 01 Crowd Moler
,~------

'0•
> -500

-'ooo~--=:-----:=----:::::---=:---:::=:---::::
o 2000 '000 6000 6000 ,0000 12000
Simulatcd Vollago ot Crowd Meler (Empincal Model)
'OOOr---~---_--'---,.---~---~---,

! SOC
g, 0
~

'0
> -500

-1ooo0.'----:20=00:---:'-=:00::0,------::6OOO=---60=00=---:,:::0000=--~'2::ooo
Snmplo Points (O.01s1point)

Figure 9.16: Actual and Simulated Voltages of Crowd Motor (Easy Digging)

• Adulli Currant of Crewd Meter

6000 '0000 '2000

51mulalod Cu""nl al Crowd Molor (Emplrical Mod.')

-400
O.'----:2QO=O,---:,c!000:::----:6000-=:::,------::6OOO=---:1-:OOOO::::::,-----:1':'!.2OOO
Sampl. PoinlS (O,O's/polnt)

Figure 9.1i: Actual and Simulated Currents of Crowd Motor (Easy Digging)

• 114
• ~
~
.t:.I

!;;;
50

0",-_"
Aetual Rolalional Angle 01 SWing Molor
15Or---r--~--~--":""'--':""'--~--~-,

100

g
~ -50
ë
c:
-,00:-----:::::::---=.:---:::::----::-:::::--:-:::::--:::::::::---:::::::::-'
o 2000 4000 6000 8000 10000 12000 14000

Simulnlod Rot&lionnl Anglo 01 Swing Molor (Empirical Modol)


15O,---~--~--""":~-""':"-~,---~---,-,
~
~'OO
.0 50
m
~
;;; 0
c
~ -50
ë
c:
-100:-----:::::::---=.:---::-:'::-----=-:::::--:-:::::--:::::::::---:::::::::-'
o 2000 4000 6000 8000 10000 12000 '4000
SDmplo Points (O.01s1point)

Figure 9.18: Actual and Simulated Rotational Angles of Swing Motor (Hard Digging)

• ~5OO
Actual Voltage of Swing Matar
looo,---~--~---.---=-~~-r--~--~-,

8-
~
>-500

-'000:-----::::.::---::::::---:::':::---::=--:::::::---==--:-:::::-'
o 2000 4000 6000 8000 '0000 '2000 '4000

Simuloled VollDgo 01 Swing Molor (Emplrical Model)


'OOOr--~--~---'---~--~-~--~-'

~5OO
8-
~
>-500

-'000:-----::::.::::---::::::---:::':::----::::::--:-:::::---:-::=-:-:::::-'
o 2000 4000 6000 8000 '0000 '2000 '4000
$ample Points (O.O'sIpoInt)

Figure 9.19: Actuai and Simulated Voltages of Swing Motor (Hard Digging)

• 115
• Aclual Ropo Position 01 Hoist Molor

o 2000 4000 6000 8000 '0000 '2000 14000

Simulatod Ropo Posilion 01 Heist Melor (Empiric..11 Modol)

o 2000 4000 6000 8000 1‫סס‬oo '2000 14000


Snmplo Points (O.Ols/poinl)

Figure 9.20: Actual and Simulated Rope Positions of Hoisl Molor (liard Digginp;)

• ~5OO
Aetual Voilage 01 Heist Meler
looor--~r-----'~--'---'---'--~---""'--'

&
m
0
'8
> -500

-1 000~-~::--7:'::----,:-:'::,------=~--:0:':"--:-::::':""-~:-:-'
o 2000 4000 6000 8000 10000 12000 14000

Simulalod vonago ot Heist Motor (Empirical Modol)


1000

-1 ooo0:----:c2OOO=---:4-:.ooo::--...,8000:::::--8000~::--:-:,OOOO:-::,:---:,2OOO:::::---:,...,4OOO~.J
samplo Points (O.Q1s/poInt)

Figure 9.21: Actuai and Simulated Voltages of Hoist Motor (Hard Digging)

• 116
• r------' ,
Aclual Curronl 01 HOlst Molor
. . , ' 1

l-looo~'~~~HH!~\~rH~ , " '" j


o 2000 4000 6000 8000 '0000 12000 "000

Simulnlcd Currcnt of Hoist Molor (EmpiriCllI Modal)

1000

-1000
0~--:2OOO~-""""000:':::---:6000::':-:""---=80:':00-:---:'''''00::00-:-'''''''' 2::000':':""-':-:.=00::-'0
$amplo Points (0.01 slpoint)

Figure 9.22: Adual and Simulated Currents of Hoist Motor (I-Iard Digging)

• Actual Arm Extonsion of Crowd Motor

-30~--:2OOO::':-:-"""'4000:':::--""6000::':-:""---=8ooo:':-:--"",""000::-:-0-""'2::ooo':':""-':-:4:::ooo::-'

Slmu\alod Arm Extension of Crowd Motor (Emplrical Modol)

-30~--:2OOO::':-:-"""'4000:':::---6000::':-:""-"""'8000:':-:---'''''OOOO::-:---'::2ooo=--'4000c::-:-l
Sampl. PeinlS (o.o'slpoint)

Figure 9.23: Actual and Simulated Arm Extensions of Crowd Motor CHard Digging)

• Hi

Simul'llod Voltngo 01 CrOwd MOIOT (Empiricnl Modo\)
1ooor---~--~--:""---~--~---'--- ,-

-1 000~-~=--"7:'::---:=--:---=-:=-----='"::"--:-::=--:-:-::~
o 2000 4000 6000 8000 10000 12000 14000
S3mplo Points (O,Ols/poinl)

Figure 9.24: Actual and Simulated Voltages of Crowd Motor (liard Digging)

• Actual Current ot Crowd Molor


6OOr-.,....---,--~--~--..,---.-----,--~----,

-400
0~--2OOO::':::---:4-:'OOO::---:6000=:-~8000=:--'::OOO=0:---:C'2OOO=:--:'-:40::00=-'

Simululed Curran! 01 Crowd Molor (Empiricol Modal)


6OOr-r-r....,.---.---~--..,---.---....,.-,--rr-....,

400
g 200

~ 0
8-200
-400
0!---:"2OOO~-~4C:OOO::-:---6000~:--"'8000~-"""10000""":--''''2OOO'':::---'::4o-000~
sample Point; (O.Ols/point)

Figure 9.25: Actual and Simulated Currents of Crowd Motor (Hard Digging)

• 118
• Current ef Heist Meter

J
Vdltage ef Heist Meter
1000

~ 0
:>
-1000
DipperTrip
500

-500

• ~

.§.
2
0

'0 50
~ 0 lJI... rl\...._rIf
cr: -50
0~-----:2:::0~00:--..l..-_"g~0~~,-lp-le-p-oi-=~-=-.00:::(B-.o-,
-sl-pe~~g:l::~:::'o-----:,-=OO==O""O.....J..----,,'-=-:!2000

Figure 9.26: Main Variables of Electric Mining Shovel

• 119

9.9 Diggability Index

Figure 9.26 includes th" main variables of th,· l'lect rie minin!, Slll)\"<·!. ln this lil':llI"'.
the typical working phase of the d"ctric mining sho\"<'1 is swin!, from a truck tl) th"
face of a muckpile (phase 1). digging t.he muckpil" (pha.,e 2). swing from the fa...·
of the muckpile t.o the t.ruck (phase a). and dumping t,hl' ore or rock into t.I\I' t.rnck
(phase 4).

Among the four phases, swing - digging - swing - dumping. exccpt. t.he dig,ll;in!,
phase, other three phases have little "ariance over different digging conditions. TIll'n'-
fore, we extract the digging phase for the hoist motor (Figure 9.27 and Fignn' 9.2~)

as well as the digging phase for the crowd motor (Figure 9.29 and Figur" !l.:lO) from
a whole shovel working phase, 1.0 invcstigate different digging conditions. ln arder 1.0
evaluate the different digging conditions, the diggability index had been est.ablislll'd
and used (Williamson, 1983, Mol 1987, and Hendricks 1989). The diggabilit.y index is
a product of the ratios of the total vertical signal variations of both motor arlllat.ur,·

• voltage and current, 1.0 the product of sampling time and t.hese signais. The !toist.
based diggability index equation is given as follows [26]:

HDI =
n.
CL IHV;+l -
n

HV;\)/CL 15 * HV;\) x
n tl

CL jH lô+1 - H Id)/CL 15 * Il Id)


1=1 i=1 i=1 i=l

= HVDI x HCDI

where n is number of reading taken during the dig cycle; IlV is the hoist. armatnre
voltage; HIis hoist armature current; 5 is sampling time; HDI is hoist based digga-
bility inde.x; HV DI is diggability index of hoist voltage; IfC DI is diggability index
of hoist current. For a comparison purposes, the same approach was used in calcu-
lating a diggability index based on crowd motor responses. The crowd motor ba,cd
diggability index equation is given as follows [26]:
Tl n n n
C DI = (L ICV;+l - CV;\)/(L \5 * CV;\) x (L IC/;+1 - C1.1)/(L \5 * CI.\)
i=1 1=1 i=1 1=1

= CVDI x CCDI (9.76)

• 120

where n is a number of reading taken during the dig cycle: CV' is the crowd armature
voltage; Clis the crowd armature current: S is the sampling time: C DI is cro\Vd
based diggability index; CV' DI is diggability index of crowd voltage; CC D1 is digga-
bility index of crowd current. Using eql1ations (9.ï.5) and (9.ï6), one can obtain the
different diggability indices shown in Table 9.1 (* - analytical mode!).

From the table, one can find that the hard digging has larger diggability indices
than has easy digging, in line \Vith actual circumstance. Since the hard digging takes
a longer time as well as the current and voltage trace>: of the hard digging are aIso
more ragged (comparing Figure 9.2ï with Figure 9.28), the absolute length of the
traces will be much longer. Furthermore, the average value of the hoist current and
voltage are lower. As a result, the equation (9.ï5) will produce a larger value. A
similar explanation applies to the crowd motor.

From the table, one can find that the simulated values are less than the actual
values. In those figures of Section 9.6, 9.ï, and 9.8, we can find that the simulated


curves have smoother shape than have the actual curves. As a result, we obtained
lower diggability indices for simulated traces.

From the table, one can also find that aIthough the diggability indices of the
crowd motor in the hard digging are larger than those in the easy digging, they are
less sensitive to variations in diggability than those of the hoist motor. This can be
explained as follows: comparing Figure 9.2ï and Figure 9.28 with Figure 9.29 and
Figure 9.30, one can find that from the easy digging to the hard digging, the traces
of the crowd motor change less than do those of the hoist motor. From Figure 2.1,
it is evident that during the digging process, the force in X s direction is normally
larger than that in Zs direction. The force in the X s direction is controlled by the
hoist motor and the force in Zs direction is controlled by the crowd motor. The
hoist motor is also more powerful (1450 H.P. @ 4ï5 volts - 2250 peak H.P.) than the
crowd motor (400 H.P. @ 4ï5 volts - 630 peak H.P.) [z6]. Therefore, The hoist motor
contributes more to the digging action. Moreover, another important efi'ect is due to
the difi'erent transmission efi'ects on the hoist and crowd motors. The crowd motor

• 121

is connected with a -V- belt transmission systcm. to transf,'r power from the crO\nl
motor to the shipper shaft gear train. Thc nct dfcct of tllC'sC ela.,tir helts is 10 ahsor!>
the high amplitude energy of the shocks from suddcn impact loads. damp il. and
gradually transfer the dissipated energy to the crowd motor and drive mechanism.
It is bclieved that the action of these energy absorptivc belts is responsible for the
relative lack of crowd motor responsiveness to variation in grollnd diggabilil.)" [26].
The hoist motor is directly connected to the gcar case. Hencc. il. orrers no absorpti"e
effects and the different digging conditions directly reflect on the hoist motor rt'Sponse.
As a result, in order to compare the hard and easy diggings. we show more tables of
the diggability indices in hoist motor (from tables 9.2 to 9. il.

• 122
r----

• L
1
/-;" ''.'//)iyyi" yi
..\ d ",,/hl! "(

.,,'i 111111 (J / t'd\ ,'a!llf


1/1\'/)/1
1

1
IJ-"I"1
-
./.).

1J.·IIO;; 1
/1('/)/

IUJïlJO
O.7K:}j
1

1
11/)/1 ('\'/)/1 ('('/)/1

O. ï:ll G 1

0.:1·1;;8 1
O.~NIO
O.Gï 1ï
1

1
O. ï:19:!

O. ï511
1

1
(' D/

0.G60!J

0.50'IG
.~ï"'llIlalt'(I\/alll('* 1 0.·19;;:! 1 O. ïG 18 1 O.:lïï:l 1 0.6018 1 O. ï58ï 1 0.'1565

// a 1'11 Diyyiu!l1 1111' D/ II(,D/I IID/ 1 CI'D/I CCD/I CD/

.-1dll"IV"llle 1 0.9:1·1:! 1.'1:!8.'5 1.:3:3-15 0.9919 1 0.8.'502 ! 0.8'1:3:3


SiulIllalnlVallle 1 o "·r·)
.·)_1- 1.1:300 0.5892 O. ï.'5:3ï 1 0.ï902 0.5956

Si7llulalct/Valuc* 1 0.585:! 1.256ï 0.ï:3.'54 0.605.'5 1 0.8168 0.'1946

Table 9.1: SUllllllary of Diggability Indices

EasyDigging2 liV Dl liCOl HDI

AelualValuc 0.9419 0.8055 0.ï58ï

SimulatedValue 0.5164 0.6100 0.:3150

• Sinlulatedl!aluc* 0..'5585

Table 9.2: Diggability Indices of Easy Digging 2

Easy Digging:3 HI/Dl


0.564i

HCDI
0.3154

HDI

Aeluall!a/ue 0.6348 0.6310 0.4006

Simulatet/Value 0.3943 0.6i15 0.264i

SimulaicdV',{-'lc* 0.402~ 0.i204 0.2899

Table 9.3: Diggability Indices of Easy Digging 3

liardDigging2 HI/Dl HeDI liDI

Aeluall/alue 0.6452 1.8i02 1.206i

Simulatcdl/alue 0.81 i4 1.0020 0.8190

SimulatedValue* 0.8625 0.9i46 0.840i


Table 9.4: Diggability Indices of Hard Digging 2

123
• lIardlJiggill.'l:1 1 Il 11) 1

.,Ie/ua/\ 'alllc\ 1.01 17


IIr'nl
1••.,-.,
) 1 .). 1
1
Il ni
Il "s'l-
..), • ;)

Sillllllalcdl 'allll' 0.5901 (].S(i!l~l 1 0.;)- l""


. J,)

Si1Jllllalcdl'alllc* 0.6:181 o ,--.)


.~. j _ --qs
1 (] . ; ) ; ) . ,

Table 9.5: Diggability Indices of \lard Dif';ginf'; :1

H ardDigging4 H\!DI IICDI fi DI

Ael ual Val lIC 1.:3166 1.5664 2.062:1

• Simulalcdl'aluc

SimlllatcdValllc*
0.467.'5

OA262
1.2.5·12

1.0062

Table 9.6: Diggability Indices of Hard Digging 4


0.586:1

OA289

H ardDigging5 flVDI HCDI 1/ DI

AetualValuc 0.7902 1.2207 0.9647

SimulatcdValuc 0.5107 1.3070 0.6674

SimulalcdValuc* 0.4970 0.9089 0.4517

Table 9.7: Diggability Indices of Hard Digging :)

• 124

Aclual Currcnl 01 Hoi::.t Molor (Ea::.y Olgglng Pha::.c)
~------r-- i
lOOOr '
1 HII 1

1-10oo~]~\rv01~WIj~
"~ "
o 100 200 300 400 500 600 700 aoo 900

Aetual Voltage of Hoist Meler (Easy Oigging Phase)


10oo.--~

" 500
~ HVi
i!:.
m
0
'0
> -soo

-1 0000~-""10::0'--:2':'00:--:300~--4":00::--500=--:600:::---::7':'00:--:a::00::---:900::
$ample Points (O.01s1point)

Figure 9.2ï: Actua! eurrent and Voltage of Hoist Motor in a Easy Digging Phase


AClunl Curront 01 Holst Meter (Hard Oigging Phaso)

1000

. '.

o 200 400 600 BOO 1000 1200

AClual Vollago of Hoist Meler (Hard 0199in9 Phaso)


'ooor---..----,---...,..----,.--~,.....--.,__,

l 500

f 0
>-500

-1000:----;=---=---:::::----:::::---::=--~=--'
o 200 400 600 BOO 1000 1200
sampi. Points (0,01s1poir.t)

Figure 9.28: :\ctual eurrent and Voltage of Hoist Motor in a Hard Digging Phase


• '000

-
Actual CUTTe"1 01 Crowd Motor (Easy OI~H1Ing
--------,~
Phas!.')
..

,
!

~
l-rr__-'-::-_-'.,-------::"-:-_~:______:"'::_---'-~~'..,-~~~--I
- '000
o 100 ~ m _ ~ _ ~ ~ _

Adual Voltago 01 Crowd Motor (Easy Diggmg Ph:lSO)


'000 ~----r-~-~_~~-'----;":"'--=---~---,-- -- -

Soo
~
0 0
~
m
<i
> -Soo
-'000 200 m 400 ~ 600 700 800 900
0 '00
Samplo Points (O.01sJpoint)

Figure 9.29: Aetual Current and Voltage of Crowd l\'!otor in a Easy Diggillg PhiL'"

• $
ë
'000
~
Actunl Curra"t 01 Crowd Motor (Hard Di99in9 Phaso)

0
~0
U
-~

-1000
0 200 400 600 ~ 1000 1200

Ac1ual Vol1ago 01 Crowd Motor (Hard Di99109 Phase)


1000

!~
&
m
0
'8
>-~

200 400 600 ~ '000 1200


Sompto Points (0.01s/poInt)

Figure 9.30: Actual CUITent and Voltage of Crowd Motor in a Hard Digging Phase

• 126

01strlbuhOn 01 Olggabihty Indices
2.5: ----,--~----r -~-~---,':':"'-'~-~-~-~---,
1
1
1

J •
hOrd4

hard3
• •
hard 1

hard2

oasy2

hardS
• )il(casy1

0.5

casy3

~ 01 U U U U M U M U
HOI (SimulO1lod Value)

Figure 9.:31: Distribution of Diggability Indices


Figure 9.:31 shows the data in Tables 9.1, 9.2, 9.3, 9.4, 9.5, 9.6, 9.ï. In this
figure, we ean sec that the diggability indices in hard digging distribute in top-right
area and the diggability indices in easy digging distribu!'e in bottom-left area. From
the distribution. we ean eonclude:

• Easy Digging BD!: 0.4 - 0.8 (aetual value)

• Easy Digging BDI: 0.2 - 0.4 (simulated value)

• Hard Digging BD!: 0.9 - 2.1 (aetual value)

• Hard Digging BD!: 0.5 - 0.9 (simulated value)

• 12ï
• Chapter 10
Conclusions and Suggestions for Future Work

This thesis has used statc-of-thc-art robotics and control th"ory 1" ill\',·s\ if',al,' t 1...
electric mining shovels as well as established a simulator for this kind ,,1' machin,·s,
The conclusions and suggestions for future work arc indi<'<1I."d in th,' foll"wing s"cli"ns,

10.1 Conclusions

This thesis has comprised thrcc main work activities: analytical modclling, cmpirical
modelling, and simulator for the electric mining shovels. Thc analytical and cmpirieal

• models have included kinematics, Jacobian, rigid body dynamics, aduator dynamics,
interaction between bllcket and muckpile, paramctric system identification, and non-
parametric system identification. Ali these models have contributed 1.0 establishing a
simulator for electric minin~~ shovc1s. The simulator was validated by using cmpirical
data. In add;;ion, because the diggability index is an important fadaI' (usually lISl·d
to evaluate the different digging conditions), the actual and simulated diggabilit.y in-
dices have becn used ta validate the simu!ator. The simulator cou Id contributc 1.0
the shove! manufacturers in their ability to improve their machine designs. The sim-
ulator could contribute to investigating the effects of different muckpile condition on
shove! performance. The kinematics could contribute to a shovcl operator's ability
to optimize their trajectory and task planning. Ali these modc1s could contribllte to
computer control design. Main achievements of the thesis are:

• An analytical mode! of the complete machine was derived, based upon rigid
body assumptions. This included the crowd, hoist, and swing motor actllator


models and kinematics, Jacobian, and rigid body dynamics.

128

• ..\11 tlllfllyt.ind lIIodd of tlw IJlICkd.-Itlllckpii(· interaction wa.." deriv{'d. Basee!
llpOrJ Howlauds' frictiotl modd of a draglitw bucket. a more ad\'élnced friction
lnocld of the "hove! hllcket \Vas ohtained. In addition. Ihe impact erfect" relevant
10 larger fragmenb of rock \Vere considered.

• In II", empirical modclling parI. system identification mcthods arc used. In


addition to a typical ARX estimate. used for identification of the transfer func-
tion", the ma:"imllm likelihood estimate \Vas employed for identification of the
"tate space modek \Vhich \Vere used in the system simulation. Furthermore, the
(ii"turbances acting on the three actuators \Vere identified by using nonparamet-
ric modcl estimation. The diggingjnon-digging, and easy diggingjhard diggil'g
effects were also considered. as governed by the fragmentation characteristics.

• An integrated modcl of the complete process, i.e. the machine and its interaction
\Vith the muckpile. \Vas constructed based on ail the previous mode!s.

• The individual models, as weil as the integrated mode!, \Vere implemented in

• software, using MATLAB and its toolbox SIMULINK, and validated by simu-
lation.

• The analytical model and empirica! models were simulated, accounting for easy
digging and hard digging conditions which were compared.

• The diggability indices of the actual and simu!ated armature voltages and cur-
rents \Vere calculated, in order to further validate the simulator.

10.2 Suggestions for Future Work

Certain improvements could be made to the modelling, simulation and analysis of the
machine:

• The models could be extended to include non-rigid body e!fects of the machine


structure.

129

• The ctnpirical tllode1~ can be r('vised by l1sing I1llln' input-lHl! pul dal" OIlId t IH'ir
identifications in order to obt.ain T11ore.' ac.:curalt' tlwd(·l:--.

• Th" control problelll conld 1", pO'<'d as a cOllplt-d Illlllti\'ariahlt- prnhit-Ill. 'l'hi,
shollld be a more accllrate r"f1ection of t h,' cOllplt-d nat nrl' of t \It' proc,'''.

• Sinec the systcn1 parmllctcrs are changing \Vith n'spl'ct t.o t.illH.'. adapt.in· CI.Hlt I",ll

could be used on the shove! control system.

• The interaction between the buckct and mnckpile Illay he ahlt, to Ill' exlt-l"I"d
to include the cohesion issue when digging a wct muckpile.

• Elements of structural design couId ~e reconsidered. e.g. using paralle! insl<'ad


of seriaI kinematics. which would greatly enhancc machine rigidity and r"duce
the disturbance loading of the actuators. while likcly reducing the weight and
size of the eiectric mining shove!.

• 130
• APPENDIX A

Appendix: Simulink Simulators

TIJ(' simulink simulators include the main routine. ~Electric Shovcl :'--Iain System".
and sorne subroutines. which are mainly three disturbance blocks. 7'.1,. Td3 • and Td.,.

• 131
G--eo
~inp;t data
[}err"Jt
CIOCk Ta lime

1'.J,'diD€r.1 EJectriC Shovel Main System

• • •

s::
c:
s::
c:
x L --;.x:.- ...J


CIl + +
c:
3
~'---r--_...J

Figure A.2: Td4 subroutine

• 133

s:
'"c.c.

s:
c:
s:
c:
x L ..2x~ -l

(")
0
ëi ëi


::>
!!1. '" '"
'"?

cn,......J~....L..,
c:
ê.L....---,---l

Figure A.:3: f43 subroutine

• 134
I Muxl ~I -O.5'u[3)'(O.5'(u[4))"2'(cos(u[2))+sln(u[2)))"2+(u[5))"2) 1 ~I+ '"
<l
.~
i ..
v4c3-v43 g
.-.1+
8uml
~
~ 10
Mux ~ .....
M

~
.;;;
<
6
'---'
th3dd ~
~

v43

• • •
.,c:
";:;
. - . l Mux ~-uI3I'u[5J-0.5'(u[4))1'2'«cos(u[2)))"2-(sin(u[2)))1'2)+0.707'(g-Ll' (u!4))1'2)'sln(u[2) "...
0
.0
v43_2
"on

-0.5'L2'(u[4))"2·sin(u[2))-(0.707'(Ll' (ul4))1'2+g)+0.5'L2'(u[4))"2)'COS(u[2))
'"~ CD

'".....
ai:'
Mux
v43_1 ..:.,
...
6'0
li:;

• • •

C1l~
0.

;::: ;:::
C C
x '- TX --J


en ,....-''-----''---,
c
3 '---:r:--'

Figure .'\.6: Td3 subroutine

• 13i
..
.
.
-.;--. ~
~
l - ..
fd - ..
}- .. Mux
. u[3]'f51
~
·..
J ·. d4f51

1 ~ 'c"
L- ';;
~ ~ ~ ...g
l .· .
.
+
+
.0

"'"
~ H

r
M
+ 00

Mux ~ O.5'u[3)'rn4 f - -
-.+- M
C .....
M

c
• 1:':
-<
..
-

.. -
Mux1
v41
U
Producl ~
"...
"
t.O

.
. ~
v4c1

·..

~
-p

- .. ~
- .. -
N42

• • •
~
0-g

~
Mux In3'u[6j+O.707'(u[4J)h 2'(coS(U[2J)+sln(u[2J))'(cos(u[2J)-sln(u[2J))'(Iyy3-lxx3) ~ ~
Z ..-<
N33_ cO
..:
.~
r=..
Mux

• • •
.,
ozc::
;;l
...
0
.a
::l
on
C'I
Mux lyy4·u(6)-O.5·(u[4))"2·((sin(u[2J))"2-(coS(U[2)))"2)·(lzZ4-1xx4) <l' 0
Z ....
<l'
N42_1
ôi
<f,
.,...
6'0
tz
Mux

• • •
.-..,c::
Q)

1Muxl ~I O.5'u[3]'u[6)+O.25·u[3)'(u[4))h2'((sin(u[2]))h2-(cos(u[2)))"2) + 1 --
i ~
v4c1-v41
2"
..0
+ 1 v4c1
....."v
<Il
Sum1 .....
~
- Mux > .....
~

Ih~
ë.....
<
'-.::1 ~
~ E'o
Ih3dd ~

v41

• • •
.,
"::J
.;:;
....0
on
::J
<Il
......
1 1 ~ 1-0.5'u[3J' (u14»"2'(cos(u[2»-sln(u!2»)"2-u[3)'(u(5»"2+0.707'(9-L1'U(4))"2'cos(u(2)) .".
>
C'l
.".
......
......
v41_2 ......
<t,
-0.5'L2'(ul4»"2 'cos(u!2))+(0.707'(Ll' (u!4))"2+9)+0.5'L2'(u!4))"2)'sln(u(2)) ~

v4C1 .-
~
É'o

Mux

• • •

Ne::;: . . .
Co

s::
c
x
~----.-----'


Figure A.12: Td1 subroutine

• 143
1
-'
th1dd
~

2
---'
lh3
3
l ~

oz"
::l
d4 Mux Izz1'u[1] ...
0
.ri
4
Izz1 + ~
th1d Sum1 C':'
.-.
5 ~
.-.
th3d
"
c:;
.-.
6
Producl ..:.,
lh3dd Mux1 ...
::l
bI)
li:;

n21
Sum

n22

• • •
L2J 1 ~I
Ihldd
2H-4 .--_~
Ih3 Mux (-L1-0.3535'l2)'m2'u[ljl 1

3 F23
d4 .,
c:l
4 H----tiH '';;
::J
Ihld g
~
~
::J
U>
Ih3d M
li:>
~
~ :1 Conslanl3 + 1 '<l'
..-.
;,;.
lh3dd 1 1 ~ Mux -O.5'l3'(O.707'u[1j'(cos(u[2J)-sln(u[2J))lI----, ..-.
L--_~
, 1 • 1
+ 1+/1 1 ..:.,
v33 ...
É'D
~
O.35'l3'u[4j'ul5]'(cos(u[2J)+sln(u[2J))-(O.707'l2+l1 )'u[1]
tJ
Sum1

51
v34

Producl1

v4c2

• • •
!J l
thldd
"1 -0.3535' u[3)'(u[1)'(cOs(u[2))+sin(u[2)))+u[4)' u[5)'(cos(u[2))-sin(u[2))))
+ 1 0
v4c2-v43 c::
o
Ih3
Mux 0.3535'u[3]'u[4)'u[5)'(sin(u[2))-cos(U[2))) + ~11
'';;

"8
.r>
v4c2-v42
3 >--1---'-'
+ 1 "
<Il

d4 '\l
....
4 r--t---H~ 1 Sum2 > '.........""
thld Muxl lr.;
.....
5 J-=t;Wt:r:-i .;,
Ih3d ...
0

"
.~
~

v42

• • •
t~d 1 :1 ~
.~
-O.707'u(3)'(u[4))'(cos(u[2))-sln(u[2))) .J:>
~
v42_ 1;'1
1-
~ .....
'<l'
O.707'u(3)'u[4)'u(5)'(Cos(u(2))+sln(u(2)))-L1'u(1)+O.707'L2'u(1)
cO
.....
v42_1 .;,

~
~

• • •
Product
---l
n31 _J M o t ..' " ' ' LJ <1
t:
1 '';;
cos ;:j

lhldd
+ ~
...
0

;:j
2 <n
Mux O.707'lxx2'u(1) + .....
lh3
3 N21
- C'I
t:
00
'<l'
......
Suml 1:'=
d4 ......
4 -<
th1d Mux sln(u[2)) ...
<1
;:j
to
~I • 1_ 1 cosl t;:;
Ih3d
~

6
-' .~

lh3dd 1 Y-r-+I YI Producl1

n32

• • •
r--

~d
J Mux ,: 0.707'lxx3'(u[ 1)'(cos(u[2))+sin(u(2)))+u14j'U[SJ' (cos(u(2))-sin(u(2))))
\ N31
J '--+'-c-
;MUK; Y .. ..
0.707'u[4j'(cos(U[2))-sin(u[2)))tu[S)'(lzz3-1yy3)
~
1
1
d
Mux r--- 1 O.707'Ixx4'(ur1j'(cos(u(2)) .sin(u[2J))tu[4)'uIS)'(cos(uI2))-sin(u[2J))) 4 +
d
1
. Nd1 + .,<:l
J '-c- ~ 0.707'u(4)'(sin(u[2))-cos(uI2J))'u(S)'(lzz4-lyy4)
1
+ '';:;
Id
. Muxs 1
1 ~
..."
0
..... N42 +

~ . .0
"0.S'm4'u(3)
],
....."
+
m4d4
'"
M 0>
Producl + <:l "'1'
..... .....
---+ W
,YiÇ.? -.j -O.S' L3'm3'(0.S'u(6))-(0.707'(g-L1'(u[4)}"2)-0.S'L2'u[4))'sin(u[2))
1
1 -.
+
+
.....
-<.,
- ...
Mux
-4 -O.S'L3'm3'(0.707'(L1'(u[4J}"2+g)tO.S'L2'(u[4)}"2)'cos(u[2))

-13/2F32
:
.. +
Sum1 ~
E'o

f----.'-c-
Mux2 !L3'm4 ~
Constant

.....
~ ~
Producl1

v4c3
1 L3'IS3 :
Conslantl

• • •
-O.S 'm3 'L3 '(O.S'(u[411"2 '(cos(u[2))-sin(u(2J)) A2+(u[SJI"2)+0.707 'm3'cos(u[2)) '(g-L 1'(u[4))'2

F32 1. ~

-0.S'm3'L2'(u[411"2'Cos(u[2))+m3'(O.707'(L1'(u[411"2+g)+0.S'L2'(u[4)))'sin(u[2))
F31 ~

"..,"'::l"
0.707 'iyy3'(u[1 1'(cos(u[2))-sin(u[2)))-u[4]'u[S)'(cos(u(2))+sln(u(2J))) f---,
1-----. L.t + 0
...
.J>
-O.707'ul4j'(cos(u[2))+sln(u[2)))'u[S}'(lzz3-1yy3) 1 1 + ::l
'"
C'I

.707'lzz4'(-cos(u[2))fsln(u(2)))+u[4]'u[S]'(cos(uI2))+sln(u[2))}---"'l-..r: '"" C>


.....
'"
.-- -N43+L3fSl ~ f---J 8um2
Ô'i
.....
.(,
10.707'U[4j'U[S]'(COS(U[2))fSin(U(2J))'(lyy4-IXX4)-L3'fSll 1
f::
i 1 -N43+L3IS2 6'D
Çi:;

v4cl

• • •
~

---.
~ ri'LJ
Product
n31
1
-.j sln(u[2)) ~
~
cos1
Id

l- ...
~
1
J
~

Mux ,;: O.707'lyy2'u(1) l -


+
III
l=l
.~

~'"
~

1
~
"l'>'> 1
+
~I
~ : O.5'l2'(-L1-0.353'L2)'m2'u[1) : - ~I

1
-L2I2F23 + ""
C'1 .....
u')
1 f-+ L--.j cos(u[2)) ~ - ë
l=l .....

l MIOV
e

~.!
..;
cos
~
~
~
~
~
~
,1
€'o
~
~

~ _---,,-J
~

1-- Product1

-. n32
l2 ~

~.-
Constant Producl2

4
133

• • •
:D 1 ~r -0.5' L3'(O.707'u[1)'(c05(u[2))-s in(u[2))))
lh1dd Il>
2
'"
oZ
.."
~

lh3 Mux 0.35 'L3'u[4)' u[5)'(cos(uI2)) tS ln (u[2)))-(0.707' L2 tL 1)'ull ) 0


3l-l--W v33
..0
d4
"
VI

~
+1 co> C'l
~ '1:'
......
~
lhld Mux4
......
~ +1 '33 <'1
..:
Ih3d
~
Ih3dd
Sum1
..
Il>

6'0
ç;:;
Producll

v4c2

fI • •
• Bibliography

[11 Anderson. W.. Controlling ElectrohydrauIic Systems, Marcel Dekker. !Ne.. 1988.
[2] Astrom. K. .J., Computer Controlled Systems: Theory and Design, Prentice-Hall,
Second cdition, 1990.

[:1] Astrom, K•.1., Introduction ta Stochastic Control Theory, Academie Press, Inc.,
1970.

[~] Belangcr, P.R., Control Enginccring Text, Fall 1990.

[5] Bollinger, J.G., Duffie, N.A., Computer Control of IHachines and Pro cesses,
Addison-Wesley Publishing Company, 1988.

• [6] Brady, RI·I.G., Brown, E.T., Rock Mechanics for Underground .I\tIining,
Boston:Allen and Unwin, 1985.

[7] Chapnik, B.V., Heppler, G.R..Aplevich, J.D., ControI1ing the Impact Response
ofa One-Link Flexible Robotic Arm, Proceeding of the International Conference
on Robotics and Automation, IEEE, 1990, pp. 1444-1449.

[8J Clarke, A.B., Disney, R.L., ProbabiIity and Random Processes for Engineers and
Scientists, John Wiley & Sons Inc., 1970.

[9J Chung, K.L., EIementary ProbabiIity Theory with Stochastic Processes Trans-
action of the Institution of Mining and Metallurgy, 1979.

[10] Colgate, E., Bogau, N., An Analysis of Cc,ntact InstahiIity in Terms of Passive
Physical Equi\alents, Proceeding of the In·.~ernational Conference on Robotics
and Automation, IEEE, 1989, pp. 404-409.

[11] Cotton, B.B., EIectrics and Hydraulics on Roadheader Machinery, IMechE, 1985,


pp. 13-22.

153

[12] Craig..1.J .. Introduction ra Robotic". '<'cond edition. :\ddison-Wes,,"y !'Ilhlishillp;
Company. 1989.

[1:3] Daneshmcnd. L.K.. and Pak. Il.:\.. Model lV-ference Adill'tin- Control of l'i'ccI
Force in Turning. Trans. :\SME..1. Dyn. Sys. ~ka". and ('ont roI, \·ul.1n~.

September 1986, pp. 215-222.

[14] Dancshmend, L., Hendrieks, C., \Vu. Il. and Scoble. M.. Design of a ~lininp;

Shovcl Simulator, Proeccdings of the International Congre,," on Mine Design.


Kingston, Ontario, Canada, 1993, pp. 551-561.

[15] Dancshmend, 1., Seoble, M., etc., 6th Canadian Symposium on jvlining Au/,oma-
tion, Montreal, Canada, Oct. lôth-19th, 1994.

[16] Davenport, W.B., Probability and Random Proeesses, an introduction for ap-
l'lied scientists and engineers, MeGraw-Hill Book Company, 19iO.

[li] D'Souza, A.F., Design of Control System, Prentice Hall, Ine., 1988.

• [18] D'Souza, A.F. and Garg, V.K, Advanced Dynamics: Modeling and Analysis,
Prentice-Hall, Ine., 1984.

[19] Franklin, G.F., Powell, J.D., and Workman, M.L., Digital Control of DYU:l/uic
Systems, Addison-Wesley Publishing Company, 1990.

[20] Goldsmith,W. Impact: The Theory and Behaviour of ColIiding Solids, Edward
Arnold Ltd., London 1960.

[21] Goodwin, G.C., and Sin, K.S., Adaptive FiItering Prediction and Control,
Prentiee-Hall, 1984.

[22] Grenier, A., Chevrette, G., and Coache, C., Noranda Automatic Guidance Sys-
tem Architecture Overview, 6th Canadian Symposium on Mining Aut.omation,
Sept.1994, pp. 116-119.

[23] Guru, B.S., Hiziroglu, H.R., EI~ctric Machinery and Transformcrs, Harcourt


Brace Jovanovich, Ine., 1988.

154

[2·11 Iladjii\""ri\ioll ..1., Slu,]i"" of A[achine - Ground Interaction in SurfilC<' Mines.
l'h.J). Il,,,sis. :-'k(;ill Cniv"rsily. 199:3.

[2;'] Il''lllarni. A.• GOlllet. S.• alld Aubertin. :'-L. On the Resist.ilnce of Particulate
Medi" 1.0 Hucket. Loading. 61h Canadian Symposium on ;\Iining Automation.
S"pt. 199·1. pp. IiI-liS.

[261 lIcndricks, C., Performance A'[onitoring of Electric Mining SlJOvels, Ph::'>. thesis,
McGill University, 1990.

[2i] lIendricks, C., Daneshmend, L., \Vu, IL and Scoble, l'vI., Design of a Simulator
for Productivity Analysis of Electric Mining Sho\'els, Proceedings of the Sec-
ond International Symposium on Mine Mechanization and Automation, Lulea
University of Technology, Sweden, 199:3, pp. :329-:3:36.

[28] lIendricks, C., Sc()ble, M., and Peck, J., Performance lvlonitoring of Electric
Mining Shovels, Transactions lnstitute ofMining and Metallurgy. 19S9, pp. A151-


A1.59.

[29J Herceg, E.E., Handbook of Measurement and Control, Pennsauken N.J.: Schae-
vitz Engineering, 19i2.

[:30] Hettiaratchi, D.R.P., and Reech, A.R., The Calculation of Passive Soil Resis-
tance, Géotechnique 24, No.3, 289-310, 19ï4.

[31] I-Iollerbach, John M., A Recursive Largrangian Formulation of Manipulator Dy-


namics and a Comparative Study of Dynamics Formulation Complexity, IEEE
Transactions on System, Man, and Cybernetics, Vo\.SMC-10, No.lI, Nov.1980.

[32] l-lostetter, G.I-I., Savant, C.J.Jr, and Stefani, R.T, Design of Feedback Control
System'_;, Saunders College Publishing, 1989.

[33] I-Iuculak, M., Compliant Motion with Impact, Time Constraints and Uncertainty,
M.Eng. thesis, McGill University, Sept. 1990.

• 155

[:>4] Izaguirre. :\.. Hashimoto. ~l.. l'au\. IL Hayward. \" .. :\ S,'II" ("'''''l'''l''li,),,,,{
Structure for Real- Timc Dy"amic". The Internati,>tlal .J,mma! of [(obol il':' H,'-
search. pp. :>46-;,6l. Ang. 1992.

[:,5] .Iohnson. C. D.. Theory of Di"l urb'lIlc'~:\CCO"llIlOd,,/.i,,~ ("Pli/. rplI.-r.-. ('Ollt roi a Il.1
Dynamic Systems: :\<h'allccs in Theory and Application:,. \"0\.12. C."I'. \.'·<>lId,·:,.
editer, Academic Press. New York. 1976.

[:36] Kahng, .J., Amirouche, F.M.L.. Impact Force Allaly"i" i" :\lecllilllicai lJ'lI,d J),.-

sign, Proceeding of the International Conference on Robotic" and Antotllatioll.


IEEE, 1987, pp. 2061-2067.

[37] Karnopp, D. and Rosenberg. R. System Dynamics: :\ Ullified :\l'l'roach. l!)7"

[38) Kauth, D.E., The TEXbook, Addison-\Vesley. 1984.

[39) Kosow, LL., Control of Electric IHachines, Englewood Cliffs. N..J.:Prentiee-llall.


1973.

• [40J Kuo, B.C., Digital Control Systems, Sri, Illinois, 1977.

[41J Kuo, B.e., Automatic Control Systems, fourth edition, Prentice-Ilall, Englewood
Cliffs, 1982.

[42) Lamport, L., LATEX: A Document Preparation System, ,\ddisoll-Wesley, 198(;'

[43) Learmont, T., Current Developments in Surface Mining Excavators, Interna-


tional Conference on Mining and Machinery, 1979.

[44) Lee, C.S.G., Robot Arm Kinematics, Dynarnics, and Control, IEEE Computer
1982, 1'1'.62-80.

[45J Lever, P.J.A., Wang, F.Y., and Shi, X., An Intelligent Task Control System for
Dynarnic Mining Environments, For presentation at the SME Annua1 Meeting,
Albuquerque, New Mexico, Feb. 14-17, 1994.

[46] Ljung L., System Identification: Theory fo~ the User, Prentice-Hall, 1987.

• 1.56

[·1i] Lj"n~ L.. .'iYS1.(,IIl Ide1ltiFication Too/box. The ?\lath Works. 1991.

[.1S! Lj"n~ L.. Sod"rst.ro,,, 'l'.• Theory and Practice of RecursÏ\'e Identification. The
:vI 1'1' Pr"ss. Cambridge. ~Iassachusetts. 198:3.

[.19] L"h ..LY.S.. Walker, ~I.W., and Paul, R.P.C., On-Line Computational Scheme
for Mechanical Manipulators, .Journal of Dynamic Systems. Nleasurement, and
Control, .June 1!J80. \'01.102, pp. 69-i6.

[50] Maciejowski, .LM., iHulti\"ariabIe Fcedback Design, Reading, Mass.: Addison-


W"sley. 1989.

[51] Ogata, K., Modern Control Engineering, Prentice-Hall, 1990.

[52] Ogata, 1":., System Dynamics, Pr ~ntice-Hall, Englewood Cliffs, New Jersey Oï6:32,
19i8.

[5:3] Palm, W.J., Modeling, Analysis, and Control of Dynamic System, New York:


WiJey, 1983.

[54] Papoulis, A., Probability. Random Variables and Stochastie Proeesses, McGraw-
Hill, New York, 1990.

[55] Patton,R., Frank,P., Clark,R., Fault Diagnosis in Dynamie Systems, Prentiee


Hall, 1989.

[56J Phillips, C.L., Nagle, H.T.,Jr., Digital Control System Analys;~ and Design,
Prentiee-Hall, Englewood Cliffs, N.J.1984.

[5i] Piche, A., Perspeeti\'es of Mine Automation in Canada, 3rd Canadian Sympo-
sium on Mining Automation, Sept.1988, Montreal, pp. 5-6.

[58] Raven, F.H., Automatie Control Engineering, 3rd edition, McGraw-Hill, 19ï8.

[59] Readman, M.C., Contributions to the Control ofCompliant-Joint Manipulators,


Ph.D. thesis. McGill University, 1992.

• 15ï

[60] Reece. A.R.. .-\ Rational :\pproach to the Design of Earth-.\/m·;ng .\/achi,ws.
.Journal of TC'rranlL'chanics. \·ol.:ll. 1\0.l. , p. li!)·,!!. 1!)~.l.

[61] Rose.D.P.. Hydraulic. Electrical. Pnellmatic COIrrroJ-\\ï,ich \1':1.1' t.. (;..:'.


Il\IechE 1985. pp. :l:3-30.

[62] Rowlands. Jeffery C.. Dragline Bucket Filling. liniversity of Qul...·nsland. l'h. n
thesis. 1991.

[6:3] Sarata, S" Concept of an autonomous system for piled ore sllOn-Jing. Mine Mech-
anization and Automation. June 199:3. Sweùen.

[64] Savant. C.J., Stefani, R.T., Design of Feedback Cont.rol Systems. N.'1'.:SallIH\c'·S
College Pub., 1989.

[65] Shinners, S.M., Modern Control System Theory and Application. Reading.
Mass.:Addison-Wesley Pub. Co" 1978.

[66] Silva, C.W.D., Control Sensors and Actuators, Prentice Hall, Englewood Cliffs,

• New Jersey 076:32, 1989.

[67] Soderquist, B.A.T., Wernersson, A., Information for Asscmbly from lmpact.s,
Prorccding of the International Conference on Robotics and Automation, IEEE,
1992, pp. 2012-2017.

[68] Spong, Mark W., Vidyasagar, M., Robot Dynamics and Control, .John Wiley
and Sons, 1988.

[69] Takahashi, Y., Rabins, M.J., Auslander, D.M., Control and Dyno.mic Systems,
Addison-Wesley Publishing Company, 1970.

[70] Taylor, D.W., Fundamentals of Soil !\lIcchanics, John Wiley and Sons, 1948.

[il] The Math Works, Inc. Simulink User's Guide, 1992.

[72] Thornhill, R.J., Smith C.C., Impact Force PredIction Using Mcasured FreCJl1cncy
Response Functions, Journal of Dynarnic SYftems, Measurement, and Control,


Dec. 1983, VoU05, pp. 227-231.

158

[i:l] Tirnosl",nko. 5., Strengt.1z of }Iaterials. :lrd edition. 1\.Y.: R.E.Krieger Pub.Co..
19i6.

[i.1] Walker, !.IL The Use of /{inema/,Îc Redundancy in Reducing Impact and Con-
tact Effect.s in Ivlanipula/,Îon, Procccdings of the International Conference on
Robotics and Automation. IEEE, 1990, pp. 4:34-4:39.

[i51 Wang, Yu, Mason, M.T., ll'Iodel1ing Impact Dynamics for Robotic Operations,
Proceeding of the International Conference on Robotics and Automation, IEEE,
19Si, pp. 6i8-6S5.

[i6J Weaver, W., Timoshenko, S.P., and Young,D.H., Vibration Problems in Engi-
neering, ,5th edition, N.Y.: Wiley, 1990.

[ii] Wu, H., Dynamic Modelling, Simulation, and Control of a Vertical Turret Tun-
neUing Machine, McGill University M.Eng. thesis, 1992.

[iSJ Wu, H., Daneshmend, L., Dynamic Simulation and Computer Control ofa Boom-

• Type Tunneliing Machine, Proccedings of the International Congress on Mine


Design, Kingston, Ontario, Canada, 1993, pp..539-549.

[i9J Wu, Il., Daneshmend, L., Dynamic Control of a Vertical Boom Type Tunne/ling
Machine, Proccedings of APCOM XXIV International Symposium on the Appli-
cation of Computers and Operations Research in the Mineral Industries, Mon-
treal, Quebec, Canada, 1993, pp. 560-56i.

[SOJ Youcef·Toumi, K., Gutz, D.A., Impact and Force Control, Proceeding of the
International Conference on Robotics and Automation, IEEE, 1989, pp. 410-
416.

• 159