2010 International Conference on Computer Application and System Modeling (ICCASM 2010)

A Flywheel Energy Storage System Suspended by Active Magnetic Bearings with

Fuzzy PID controller

ZHANG Weiwei
School of Mechanical Engineering, Ningbo University of technology, Ningbo, China, 3 150 16
e-mail: zdzdq@ 126.com

Abstract-A flywheel energy storage system is developed. The The disk-shaped rotor measures 302 mm in outer diameter
structure and dynamics characteristic of the system are and 156 mm in inner diameter. The rotor mass is 1.0 1kg,
discussed. The system consists of a disk-shaped rotor, active which is supported by attractive force produced by three
magnetic bearing (AMB), PIO controller, displacement sensor active magnetic bearings. In order to suppress the rotor
and cabinet, etc. The rotor is suspended by three active vibration, PID control is applied to the AMB.
magnetic bearings (AMB). A mathematical model of the
system is derived and each degree of freedom motion is treated
separately for the control system. PIO control is applied to the
AMB. The experiments have been carried out to measure the
dynamic response of the rotor to direct disturbance. The
results indicate that the fuzzy PIO controller possesses good
performance.

Keywords- flywheel; energy storage system; active magnetic

bearing; Fuzzy PID control

I. INTRODUCTION

Recently saving energy effectively has become very

important. Thus, energy-saving technology for surplus power
is very important. A flywheel energy storage system is one Figure 1. Schematic illustration of the flywheel energy storage system.
of the valuable energy-saving technologies. The system is
very simple and characterized by high energy saving B. Model
compared with other energy storage systems. They are one translational motion in the vertical direction
In this paper, we propose a prototype of flywheel energy e
storage system. The rotor is suspended by three active z and two rotational motions, roll x and pitch By . The
magnetic bearings CAMB) to constrain three degrees of coordinate system and the mathematical model of the system
freedom along z-axis and about x-axis and y-axis. The other are shown in Figure 2.
three degrees of freedom about z-axis and along x-axis and
y-axis are to be passively constrained by edge effect[J].
Active magnetic bearings are ideally suited for high-speed
and vacuum applications due to their contact free operation,
low friction losses, adjustable damping, and stiffuess
characteristics and due to the fact no lubricants are
necessary. During normal operation, there is a 12.5mm air (a)
gap between the disk-shaped rotor surfaces and the stator.
PID control method is applied to the flywheel energy storage
system. The dynamics of the energy storage system is
discussed.

II. SYSTEM

A. Structure (b) y (c)

We have developed a prototype miniature flywheel
energy storage system. The schematic illustration of the Figure 2. Mathematical model of the flywheel energy storage system.

flywheel energy storage system is shown in Figure 1[2J.The

system consists of a disk-shaped rotor, active magnetic It is assumed that the roll and pitch angles are so small
bearing, PID controller, displacement sensor and cabinet, etc. cosex :::::l,sine x=ex' sinev =ev cosev = 1,
that

978-1-4244-7237-6/10/$26.00 2010 IEEE V5-116

 2010 International Conference on Computer Application and System Modeling (ICCASM 2010)

where ex ,Bv are the roll and pitch angles of the rotor. The
then writing
I fz d l
[] []
z Zb
:: , d = l:::
motion equations of the rotor can be written as
q= ,qb= F

[M] ( I)
J
From equations(1 )to(5),the dynamics equation of the
system can be written as
1.1

Mq=k,ABCq+kxACq+kxACqb+Fd (6)
Where
Where
M=diag(m, Jx, Jy) (2)
m is the mass of the rotor; 1 1 .J3 1
Jx
--a --a
is the inertial momentum of the rotor about x-axis; 2 2
Jy is the inertial momentum of the rotor about y-axis; A= 0
.J3a .J3a
,C = 0 a
fz is the z-direction electromagnetic force produced by the 2 2
a a .J3 1
three active magnetic bearings; a -a --a
mx is the moment about x-axis by the three active magnetic 2 2 2 2
bearings; B is decided by the control strategy.
my is the moment about y-axis by the three active magnetic
bearings; III. ACTIVE Fuzzy CONTROLLER DESIGN

fzd is the z-direction direct disturbance;
m/ is the direct disturbance about the x-axis;
m/ is the direct disturbance about the y-axis;
Three magnetic bearings are located at the vertices of an
equilateral triangle. The total force and torques are related
with the electromagnetic forces generated by the three active
magnetic bearings as
1 1
It is important to choose suitable control strategy and
control parameters. The flywheel energy storage system is a
typical nonlinear system, so the common linear PID control
strategy isn't suitable. An active fuzzy controller is
developed in this paper.
Fuzzy control strategy doesn't need accurate
mathematical model of the system to be controlled, which
possess strong robustness and isn't sensitive to the variation
of process parameters.

r
.J3a a
0 A. Fuzzy o/the input and output variables
2
- The displacement error e of the disk-shaped rotor and the
a a ;;
a - change rate of displacement error ec are taken as the input
2 2
-
variables. The input variables are divided into seven grade
(3) and defined as NB(negative big), NM(negative
where middle),NS(negative small), ZE(zero), PS(positive
fk is the attractive force of the active magnetic bearing small),PM(positive middle) and PB(positive big).
k(k=l, 2, 3) and a is the distance of the active magnetic The displacement u of the disk-shaped rotor is taken as
bearing from the center. the output variable of the fuzzy controller, which is also
The force of each active magnetic bearing is divided into seven grade and the linguistic variables are the
approximately given by same as input variables.
(4) Defining NB=-3, NM=-2, NS=-I, ZE=O, PS=I, PM=2,
where PB=3. Trigonometric function is chosen as the membership
Xk is the deviation of the gap between the electromagnet function of input and output variables shown in Figure 3.
and the rotor, ik is the control current. The gaps are related
with the displacements as B. Fuzzy control strategy
The developed flywheel storage system is a three-channel
multiple-input multiple output system. The coil current of
each magnets is controlled on the basis of local information
at the corresponding position as shown in Figure 4. From
Figure I, it can be seen that the positions detected by sensor
(5) I, 2, 3 are not the electromagnet gap, which requires to be
The relationships between i1, i2, i3 and x], X2, X3 are transformed in to the displacement signals of the magnets
decided by the control strategy. It can be written as respectively.

[l [:J B

V5-117
 2010 International Conference on Computer Application and System Modeling (ICCASM 2010)

:<Be NMe NS.' zOe PSe PMe PBe

NB<-' PB<-' PBe PMe PMe PSe Zo.' zOe

TABLE I!. CONTROL STRATEGY TABLE OF KI

Figure 3. Membership function

sensorl controllerl
:<Be NMe NSe NSe zOe PSe PS.
zOe
PS.
PMe Zo.' zOe PS. PS. PMe PBe PBe
zOe zOe PS. PI" PMe PBe PBe

TABLE IlL CONTROL STRATEGY TABLE OF KD

Figure4. Diagram of local control.

PBe
Controller is the core link to achieve stable levitation of
the flywheel storage system, thus the real time accuracy and :<Be PSe PS.

stability of the system should be fully considered. According

zOe
to (6), a decoupling fuzzy PID control system is designed as zOe zOe zOe
shown in Figure 5,where J is the proportional coefficient, PS. zOe zOe zOe zOe zOe zOe zOe
K; is the integral coefficient and Kd is the derivative PI. PBe NSe PS.' PSe PSe PSe
PBe
coefficient. PBe Pl+' PMoJ PloJ PS.:' PSoJ PBe

IV. EXPERIMENTS AND DISCUSSIONS

The schematic diagram of the manufactured flywheel

storage energy system for experiment study is shown in
Figure5. The area of the magnetic pole is 2.52 X 1O_4m2,
single-pole turn 2400, bias current 2.5A, offset position
12.5mm. Half-bridge switching power amplifier is adopted
NO
to provide 120V DC strong power-supply voltage. The gap
between the rotor and the stator is measured by eddy current
displacement. Conversion ratio of 12bit AID is 60 J1 s.
Figure 5. Block diagram of the control system. Sampling rate of 12 bit is 100 J1 s and the control cycle is
approximately 1ms.
According to the experience in practice, the error e and
the change rate ec should follow the rules:
ll
I)While e is lager, Kp should be adjusted to be a lager
value to improve the response of the system, Kd and K; all
should be adjusted to a smaller value;
li
2) While e is smaller, J and K; should be adjusted to be
a lager value to improve the stability of the system, Kd

l l
depends on ee .
According to the above rules, the control strategy of Kp,
K; and Kd can be obtained shown as Table 1, Table 2, and
Table 3.
Figure 6. Flywheel energy storage system.
TABLE! CONTROL STRATEGY TABLE OF Kp

V5-118
 2010 International Conference on Computer Application and System Modeling (ICCASM 2010)

When the disk-shaped rotor was suspended from floating suspension after 1 second, the gap between the displacement
to stability, the displacement curve is shown as Figure 7. sensor and the disk-shaped rotor is 8mm smaller than that of
It can be seen that the system possessed good the electromagnet and the rotor.
performance with fuzzy PID controller and achieved stable

1 1 1
200 ------ - 1 -------------
1 , 1 1

:;,
'6
180

160
---

- - --
1
1
- - ---
,
fuzzy PIP con rol 1
- - - -
---- -
1

'6
250 1 T

1 1
m

'0 140 - - - - ,- - - - -
15 T '0
" 200


c
m 120 ., - - -
E T

:6 100
m 1 1

I
'6
150

- - - -- ; - - -
1
- -,.. - - - -1-- - -

:6
"-
1

'6 80 - - - -l- - - + - - ----
1 --1----
m 1
1I11l_UZ;<V PIDlcontrol


100
1 -------------
60 ... 1

50 40 50
100 200 300 400 500 0 300 400
sampling do1JO.01s sampling dotlO.01s sampling doUO.01s

(a) PID (2.2, 0.001, 30) (b) PID (24, 0.001, 30) (c) PID (1. 6, 0.001, 30)

Figure 7. Response curve of the disk with chirp exciting signal

IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. II,

V. CONCLUSION
A prototype of flywheel energy storage system for
experiment was developed. Fuzzy PID controller was
successfully applied to suspend the rotor under outer
disturbance conditions, which possessed good dynamic and
static properties. The established mathematical model can
accurately reflect the characteristics of control. The
experiment results revealed that the performance of the
system under fuzzy PID controller is better than common
PID controller.
REFERENCE
[ I) Md. Emdadul Hoque, Masaya Takasaki, Yuji Ishino, Development of
a Three-Axis Active Vibration Isolator Using Zero-Power Control,
NO.4, AUGUST 2006, pp. 462-470.
[2) M. A. Pichot, 1. P. Kajs, B. R. Murphy, Active Magnetic Bearings for
Energy Storage Systems for Combat Vehicles, IEEE Transactions on
Magnetic, 2001,1,(37), pp. 318-323.
[3 ] Xiaoguang Wang, Kui Jiang, Study on the Centripetal Effect of the
Magnetic Bearing, the 2th Chinese Symp. on Magnetic Bearing,
Nanjing, China, August, 2007, pp. 195-198.
[4] Yefa Hu, Zude Zhou, Zhengfeng Jiang. , The Basic Theroy and
Application of Active Magentic Bearing, Beijing China Machine
Press,2006.
[5] Selim Sivrioglu, Kenzo Nonami, Active Permanent Magnet Support
for a Superconducting Magnetic-Bearing Flywheel Rotor, IEEE
transactions on applied superconductivity, Vol. 10, NO. 4, December
2000, pp. 1673-1677.

V5-U9