Vous êtes sur la page 1sur 8

Fuzzy Logic Based Soft-start for Induction Motor Drives

R.S. Janardhana Iyengar and

V.V.

Sastry, Sen. Mem. IEEE

Dept of Electric(a1 Engineering Indian Institute of Technology. Madras - 600 036 , INDIA.
In recent years, fuzzy logic has receiv,edgreater emphasis in the field of power electronics and motion control by virtue of its adaptive capability. A novel fuzzy logic based soft-start scheme for induction motor drives close {CO optimal performance has been discussed here using microprocessor based thyristorised voltagta controller. Rule-based soft-start algorithm is fully realised through a software approach only. The soft-start strategy is based on the rise in the power drawn during starting period and the sudden fall in its magnitude at the end of soft-start has been used conceptually in closed-loop. The prototype has been tested under various loading conditions and found to be reliable and near optimal.

Abstract

across the motor. The current limit is adjusted successively tilt1 the motor is accelerated.
I1 SYSTEM HARDWARE

The experiment.al set-up shown in Fig.1 is a microprocessor based voltage controlled induction motor drive used for the above scheme. Fig.2 shows the mic~roprocessorsystem with input and output ports and timer, being interfaced with the necessary instrumentation system for sensing required parameters from the point of control strategies. Fig.3 shows 12-pulse 20 width triggering scheme used to trigger the thyristors. Fig.4 shows the current (I) sensing and Fig.5 shows Icos4 sensing circuits.

I INTRODUCTION

Electric drives are the heart of any industry. Highly reliable soft-starts for motor with energy saving aspects are very much required, owing to their importance [ l l , [21 , [31 . Soft-start schemes hitherto reported [ 5 , 6 , 7 1 show the gradual increase of voltage across the motor, so that the rated voltage will appear across the motor at the end of soft-start, thus increa,singstator copper losses and undesirable harmonics at least at light loads. It was observed that the efficiency was low and power factor was poor in that kind of soft-starts under no load and light load conditions. Since the system goes to the verge of full conduction of thyristors, there is a loss of control. Some times the drive system leads to oscillations [ 8 1 . The novel soft-start schemes reported in [ 9 1 have eliminated the above problems of the conventional soft-starts and they apply close to optimal voltage at the end of soft-start, thus maintaining high efficiency, better powerfactor and reduced copper losses. The design of novel soft-start schemes depends on three input parameters, viz., (i) current limit (CTJRLIM) , (ii) delay time for varying CY by ACY (DTAcY) and (iii) Maximum firing angle (a,, ) , which are set through three sets of dip switches, SW1, SW2 and SW3. Fuzzy logic controllers have become very popular in various engineering and industrial applications and likewise in the field of power electronics and drives also. The superiority and robustness of fuzzy controllers over conventional controllers for speed and position control of DC and AC drives are illustrated in
[10,11,121. H o w e v e r
the

-.__

Fig. 1. Voltage Controlled Induction Motor D r i v e .

application

of

fuzzy

logic for soft-starting of induction motors is not reported in the literature. The proposed fuzzy logic soft-start has adaptive and selfcommissioning capabilities, thus eliminating dip switches. It chooses the parameters on which it is designed on-line and starts the motor smoothly by applying minimal required voltage
0-7803-:3008-0195 $4.00 0 1995 IEEE
121

/ellNI
AHPL

Fig. 2.. Microprocessor System.

M on ostable

Fig 5(a) 1 Cos Q sensing C K t

5v

RN

t ---

At v z
L

I
1

lil
1
I

_-

A I tCor vp

&II
1

1I S A t the output ol l H CKt

Fig.5 (b) ,Waveform showing

Cos Ct, measurement

I11 DESIGN OF FUZZY LOGIC BASED SOFT-START

SCHEME

"X

-*-

lo Channel
2
cl A D C

meaSurtnlerit

Fig. 4 (a).

Fundamental current sensing circuit

d c

F,g. 4 ( b ) , W a v e f w m a t different poiinls c f F u n d c m e n t u ! filter

The design procedure of novel soft-starts is based on three parameters, viz., (i) CURLIM, (ii) DTAa and (iii) amax. These parameters are set manually with the help of DIP switches, quite approximately, without knowing the dynamics of the system. Some times they are set by trial and observation method as well. In the fuzzy logic based soft-start scheme these parameters, viz., amax, DTlAa, CURLIM and DT2Aa are set by the controller itself depending on the dynamics of the system in terms of a set of fuzzy rules. amax and DTlAa are decided by I C O S ~ and A I c o s r ) . In the universe of discourse there are five subsets for both I C O S ~ and AIcos4. apax has four subsets, current limit (CURLIM) is decided by I and AI. The number of subsets depends upon the number of fuzzy rules required to derive the required parameters. Whatmore, the number of subsets in the independent parameters certainly improve the accuracy of the fuzzy system. The linguistic terms used for membership functions are : PVB (Positive Very Big), PB (Positive Big), PM (Positive Medium), PS (Positive Small), ZE (Zero), NS (Negative Small), NM (Negative Medium), NB (Negative Big) and NVB (Negative Very Big). I and AI both have seven subsets. The delay for moving a by ha during starting is DT2Aa is decided by E(k) and (k) - Ices$ (k-1) and CE(k) wherein E(k) = I C O S ~ CE(k) = E ( k ) - E(k-1). Both E and CE have nine subsets. DTlAa is used while decrementing a from amax to (amaX 4 O ) . DT2Aa is used while decrementing a in the closed-loop during
122

starting, and its value keeps on changing in each pass depending on the magnitude of E and CE, thus making adaptive, taking in to account the dynamics of the system.
Rules f o r , a setting : Fig.6 shows the membership function for Icos$, AICOS$, amaxand DTlAa. Table 1 shows rule matrix for cymax. the rules are on the following lines : If (ICOS$I= PS AND AICOS$ = PS) THEN amax = PS. If (Icos4 = PM AND AICOS$ = PS) THEN CYmax = PM. etc., Rules f o r DTlAa setting : Fig.6 shows the membership function for DTlAa. Table 2 shows rule matrix for DTlAci setting, where For Ices$ : PS = OlAH, PM = 034H, PB = 04EH and PVB = 072H. For AIcos$ : PS = 012H, PM = 02AH, PB = 040H and PVB = 058H. For cyinax : PS = 108O, PM = 112O, PB = 114O and PVB = 118' For DTlAol : PS = 0.3O/1 cycle, PM = 0.3O/3 cycle, PB = 0.3O/6 cycle, PVB = 0.3O/10 cycle.

Rules f o r CURLIM : The membership function for I and AI and also for CURLIM are shown in Fig.7. This i s the initial value of CURLIM taken in the beginning of soft-starting. Table 3 shows the rule matrix.

Table 3). Acter evaluaEing a,,, DTlAa and initial current .Limit CURLIM the pulses are inhibited and the soft-start process is reinitiated with the CY, The thyristors are fired with C Y , , , and a 8ilay of 9 cycles is allowed to let the current to build up. Then CY is decremented from amaxto (cymax - 4O) as per DTIAU to get initial current ramp. Then the soft-start is achieved according to the closedloop configuratiori shown (vide Fig.9) based on error E and change of error CE (flowchart vide Fig.10). In each pass of the closed-loop operation, the current is compared with CURLIM and based on the magnitude of I and AI, a decision is taken to fix new CURLIM as illustrated in the flowchart shown in Fig.11. The decrementiiig of CY in the closed-loop is done as per D T ~ A c Y . Since E and CE depend on the variation of Icosq5 the value of D T ~ A c Y may change, following the dynamics of the system as per Table 4 in each pass. The end of a softstart can be identified when E becomes negative (of any nature and it is possible to come out of soft-start at any required point). In the present work the end of soft-start is defined when E becomes NM.

For I For A I CURLIlvl

-70H, NS = -60H, ZE = OOH, PS = 60H, PM = 70H, PB = 90H and PVB = AOH. NM = -20H, NS = -10H, ZE = OOH, PS = 20H, PM = 30H, PB = 30H and PVB = 50H. PS = 60H, PM = 80H, PB = 90H and PVB = AOH.
NM =

tJ I cos P

t
Mcmbershlp function lor I
CO5

The selection of initial current limit CURLIM depends on I and AI in the following manner. If (I = Xl AND AI = yl) THEN CURLIM = zl. IF (I = ~2 AND AI = y2) THEN CURLIM = 22.
Rules

+PAICOS

Altos Q

--+

DT2Aa : Fig.8 shows membership function for E, CE and DT2Aa. Table 4 shows the rule matrix for DT2Acu. D T ~ A c Y is evaluated in each :pass of the closed-loop operation based on error E and change of error CE.
for

Membership function for A i C O S P

IV IMPLEMENTATION

The fuzzy soft-start scheme is implemented according to the schematic representation as shown in Fig.9. This is a closed-loop soft-st,art scheme. The flowchart given in Fig.10 highlights the realisation of fuzzy soft-start scheme through a software approach. To start with an experimental test is performed by firing the thyristors at test CY (116O).This is required to get value of amax, D T ~ A C Yand initial CURLIM. the test procedure is as follows: Initially thyristors are fired at test CY of 116O, and then I and I C O S ~ are read with ADC. Then cy is decremented by 2O in steps of 0 . 5 O per cycle. Once again ADC is read for I and I C O S ~ ,and based on the present and previous values of I and I1zosq5, AI and AIcosq5 are evaluated. With the values of I C O S ~ , AICOS$ as shown in flowchart both cymaK and DTlAcv are evaluated using the linguistic inferences and following the rule matrices (vide Tables 1 and 2). Similarly the initial current limit CURLIM is evaluated based on I and AI following the rule matrix (vide
'123

CCmOX

--+U

Membership turiciicn f o r

DT~AK

Fig.6 Mombeiship

function for D l 1

A N

Membership junction for a furzy loglc based soft-rturt ( t o d ~ k r m i n e ~ DT10.c) ~~x,

PI

TABLE
Fu::y

1
(Imrri

r u l e 1~8aT.r:::f o r

settiny

NS

NL

I Membershlp function torcurrent 1

pn

--

PS
pn

PPI
pn pn

PB PB

PVB PVB

PB PVB

---

PM

PB

PVB

Ai

M e m b e r s h i p f u r i c t l o n f o r 61
TABLE 2
Fuzzy r u l e matrix f o r D T l A a

-.

CURLIM

MQmberchip f u n c t i o n for C U R L I M

PPI

----

PM

pn

PB PB PVB

PB PVB PVB

Fig, 7 , &emberslip furctions f o r o fuzzy logic bnsed s o f t - & t o r t I t o determino CURLIM 1

PB PVB

PB PVB

PB PVB

VE
PL

P5

PM

PB

AI
E
M e m b e r s h i p t u n c i l o n f o r error E

\
m
NS

NM

NS
PS PS

ZE
PS PS

PS
PM

PM

PB PE PM

PVB

PS PS

pn
P I I

en
PB

pn

CE
M e m b e r s h i p tuvctlon tor c h o n g c of error CE
TABLE 4
Fuzzy r u l e matrix for OTZAa setting

NV0 NVB
NB
. .

NB
_-

Nn PVB

NS PVB
PVB

ZE
PM

PS PS es PS PS

P I 7 PM

PB PB
PB

PVB PVB
PVB

~-

._ PYB

en.
PM pn

en
PM

DT 2 A a MQmbershlp t u n c 7 i o n for D T Z A X

m
NS

__
. .

__ __

PVB

PVB

PB

PVB

PVB

PVB

PM

PB

PVB

Fig.. 8 Membership fundicns b r a fuzzy Logic based soft-start ( to deknaim DTZD 6 )


E

124

a . 3 d

Controller

Fig,9.Faz z y lcytc based soft-start scheme

[ R e a d ADC for I cos

6, __ l e t i t he c

I c o s 9 2 , find A l c o s

I call

C U R C H K fur Current iimit

1
1

1
F a d ADC for I c o s @ let i t be 1 c o s x 2

select suitably w w x , and D T l A -

.--:>
or NHorNB

1
1 c o s x 1 = 1 c o s x2 Elk-I1 = E ( k )

B u s e d on the value o f CE select suitably Dl 2 A K

I L-.--L

Fig.lO,F!owchart

for f u z z y s o f t - s t a r t

scheme

125

Fig.ll.Flow

chart b f i x CURLIM in fmzy &-start

during stcirting

ilme i n

secs.

(b)

,I

V EXPERIMENTAL RESULTS

Fuzzy logic based soft-start is tested on a 2.2kW, 4 0 0 V , 5 0 H z squirrel cage motor for different loading conditions (with a fan type of loading arrangement) as shown in Figs. 12, 13 and 14. Each of these figures show the following curves : (a) Starting current - time, (b) I C O S ~ - time, VminSeq - time and Vso rce - time, (c) CT current - time and (d) SpeeJ - time. The step variation in current shows that the current limit is automatically adjusted in steps till the motor has accelerated. The variation Ices@ also follows the variation of current. There is a small step variation in Vminreq applied across the motor indicating that some amount of time is allowed for the motor to develop the back-emf, so that the sudden rate of rise of current is controlled thus reducing the voltage dip in the power system. If we observe the speed - time curve, there is an initial delay and afterwards speed rises sharply, thus starting the motor. A close observation of the above curves shows that Fig. 12.Exger1mental waveforms for fuzzy logic based soft-start the time required to start the motor on load will be more than that on no load. Thus the TL = 0 . 5 " time required to start the motor on T, = 0.5 Nm is 10 secs, with T L = 2.5 Nm is 12.5 secs and with T, = 7 . 5 Nm is 17.0 secs.

126

:
I

11 d v = 1 1 0 volts

... R
U
U

'

p .

i
1 dlv

Time i n
540

.secS.

20
7

1 3
1

rpm

L
0

'

Time i n

1
Secs.

(4)

20

Time

secs

(d)

20

Fig.13.Experimental waveforms for fuzzy Logic based soft-start

Fig.14, Erperiitiental waveforms for fuzzy logic based soft-start

1~=25Nm

TL = 7 , 5 Nm

VI CONCLUSIONS

REFERENCES

This paper presents a novel fuzzy softstart scheme based on input power measurement. This izcheme is superior over other schemes since it has on-line adaptive parameter setting capability and also the ability to start any motor at any load, without any dip switches. Experimental results are presented to validate the above concept. The fuzzy based soft-start for induction motor drives does not call for a commissioning engineer.

Sylvester J.Campbel1, "SOLID STATE AC MOTOR CONTROLS", MARCEL DEKKAR, Inc., New York and Basel, pp. 27 - 77. Alexander S.Langsdorf, THEORY OF A L T E R N A T I N G C U R R E N T MACHINERY", McGraw-Hill, Inc., New York, 1995, pp. 245 - 290. H. Aragon, Joseph Nevelsteem, "Starting of large Motors Methods and Economics", IEEE Trans. on IA, Vo1.25, No. 6 , Nov/Dec 1988, pp.
1012- 1018.

127

T .M . Rowan, T .A. Lipo, "A Quantitative of Induction Motor Analysis Performance Improvement by SCR Voltage Control", IEEE Trans. on IA, Vo1.19, No. 4, JulyjAug 1983, pp. 545
- 553.

T.R. Gururaja Rao, "Smart Induction Motor Controller with Software Diagnostics", M.S.Thesis, IIT, Madras, 1989. G.Bhuvaneswari, "Intelligent Induction Motor Drive", Ph.D Thesis, IIT, Madras, 1992. G. Bhuvaneswari, V.V. Sastry, "Energy Efficient Solid-state Induction Motor Controllert1,European Conference on Power Electronics and applications, 3-6, September 1991, Florence, Italy, pp.2.359-2.362. R.S. Janardhana Iyengar, V.V. Sastry, "Oscillations in Voltage Controlled Induction Motor Drives", International Conference on Power Electronics and Drive system PEDS' 95, 21-24, February 1995, Singapore, pp.123-128. Janardhana Iyengar, "Novel R.S. Techniques for Energy Efficient Voltage Controlled Induction Motor DrivesIt, Ph.D Thesis submitted to IIT, Madras, May 1995. R. Johnston, "FUZZY LOGIC CONTROL", GEC Journal of Research, Vol.11, N0.2, 1994, pp.99-109. S .S.Min, K.C. Lee, J.W.Song and K . B . Cho, "A Fuzzy Current Controller for Field-Oriented Controlled Induction Machine by Fuzzy Rule", pesc' 92, 23rd Annual IEEE Power Electronics Specialists Conference, Toledo, Spain, pp.265-270. G.C.D.Sousa and Bimal K.Bose, "A Fuzzy Set Theory Based Control of a Phase-Controlled Converter DC Machine Drive", IEEE Transactions on Industry Applications, V01.30, No.1, January/February 1994, pp.34-44. M.M. Mahesh, K.P. Mahadevan and M.Chidambaram, "Fuzzy control of a semi-batch copolymerization reactor", Chemical Engineering and Processing, V01.32 (1993), pp.327-331

128

Vous aimerez peut-être aussi