Anda di halaman 1dari 7

International Journal of Automation and Computing

05(4), October 2008, 406-412


DOI: 10.1007/s11633-008-0406-3

Position Control of Synchronous Motor Drive by Modified


Adaptive Two-phase Sliding Mode Controller
Mohamed Said Sayed Ahmed

Ping Zhang

Yun-Jie Wu

Automation Science and Electrical Engineering School, Beijing University of Aeronautics and Astronautics, Beijing 100083, PRC

Abstract: A modied adaptive two-phase sliding mode controller for the synchronous motor drive that is highly robust to uncertainties and external disturbances is proposed in this paper. The proposed controller uses two-phase sliding mode control (SMC) where
the 1st phase mainly controls the system in steady states and disturbed states it is a smoothing phase. The 2nd phase is used mainly
in the case of disturbed states. Also, it is an autotuning phase and uses a simple adaptive algorithm to tune the gain of conventional
variable structure control (VSC). The modied controller is useful in position control of a permanent magnet synchronous drive.
Keywords:

Reaching law, adaptive control, sliding mode control (SMC), chattering.

Introduction

Sliding mode control has been widely used in recent


years. The theory of sliding mode control (SMC) or variable
structure control (VSC) is based on the concept of varying
the structure of the controller by changing the states of a
system in order to obtain a desired response. Generally, a
switching control action is adopted to switch among dierent structures, and the system state is forced to move along
the chosen manifold, called the switching manifold which
determines the closed loop system behavior[1, 2] . In recent
years, considerable eorts have been made to study the concepts of SMC design[3, 4] . VSC with sliding mode is a special
control technique capable of making a control system very
robust with respect to system parameters variations, external disturbances, and fast dynamic response. In addition,
the technique provides an easy way to design the control
law for linear or nonlinear plants. The VSC has a wide
range of applications such as robot control, motor control,
aircraft control, spacecraft control, inertial platform, process control, and so on. Important improvements have been
made in VSC systems, including control of the performance
of the reaching mode and reduction of chattering. Chattering occurs when the control input switches discontinuously across the boundary. This is undesirable because it
involves high control activity and will excite high frequency
dynamics, thus damaging the electrical systems and causing
high vibration in mechanical systems. To eliminate chattering, various methods are used such as the continuation
method[1, 2] , the fuzzy, neural sliding mode controller[3, 4] ,
sliding mode controller with sliding sector[5] , the reaching
law method[6, 7] , sliding mode with unied smooth control
law[8] , dynamic SMC and higher order sliding surface[9, 10] .
Lin et al.[11] used integral action and adaptive algorithm to
estimate the bound of uncertainties to reduce the switching
gain. This work is based on their algorithm.
In this paper, a modied adaptive two-phase variable
structure control is presented to control the chattering problem and keep the robustness of the systems. This borrows
Manuscript received November 21, 2007; revised April 10, 2008
*Corresponding author. E-mail address: zhp@buaa.edu.cn

the idea from the two-phase variable structure control where


the 1st phase is a smoothing phase and mainly controls the
system in steady states while the 2nd phase is autotuning
gain and used mainly in the case of disturbed states. Also,
a systematic design procedure of the modied controller is
presented and applied to a permanent magnet synchronous
motor (PMSM) to control the position of the motor.
This paper is organized as follows. In Section 2, a dynamic model of permanent magnet synchronous motor is
presented. Section 3 addresses the VSC with sliding mode
controller and chattering in VSC systems. In Section 4, the
theory of adaptive SMC and the modied reaching law are
described in detail. In Section 5, the design of the modied
controller is applied to PMSM and the simulation results
are presented. Finally, some conclusions are drawn in Section 6.

Permanent magnet synchronous motor

The permanent magnet synchronous motor (PMSM) has


numerous advantages over other machines conventionally
used for AC drives. To achieve fast four-quadrant operation and smooth starting and acceleration, the eld-oriented
control (FOC) or vector control is used in the design of the
synchronous motor drive.
However, the control performance of the PMSM is inuenced by the external load disturbances, unmodeled and
nonlinear dynamics and parameter variations such as the
stator resistance, ux, and inertia due to the temperature
rise and load variations.
The following assumptions are made in the derivation[12] .
1) The induced EMF is sinusoidal; 2) Saturation is neglected although it can be taken into account by parameter
changes; 3) Eddy currents and hysteresis losses are negligible; 4) There are no eld current dynamics; 5) There is no
cage on the rotor.
With these assumptions, the stator d q equations of the
PMSM in the rotor reference frame are as follows:
uq = Riq +

dq
+ s d
dt

(1)

Mohamed Said Sayed Ahmed et al. / Position Control of Synchronous Motor Drive by Modified

dd
s q
dt

407

(2)

where fe is the static friction, fc is the Coulomb friction,


and s is the Stribeck velocity.

q = Lq iq

(3)

d = Ld id + Lmd If d = Ld id + af

(4)

s = p r

(5)

ud = Rid +
and

where ud and uq are d q frame stator voltages, id and iq


are d q frame stator currents, Ld and Lq are d q axes
inductances, d and q are d q stator ux linkages, af is
the ux linkage due to the rotor magnets linking the stator,
R is the stator resistance, s is inverter frequency, If d is the
equivalent d-axis magnetizing current; Lmd is d-axis mutual
inductance, P is the number of pole pairs, r is rotor speed.
The electric torque is
3p [af iq + (Ld Lq )id iq ]
Te =
.
2
The equation for the motor dynamics is
dwr
Te = TL + Bwr + J
dt

(6)

(7)

(8)

3P Lmd If d
.
2
Then, we have the state equations of the motor
Kt =

d
= r
dt

(9)

(10)

B
dwr
Kt
TL
= r +
iq
.
(11)
dt
J
J
J
The PMSM drive system can be simplied to the control system block diagram shown in Fig. 1. By adding the
frictions model which is equal to the static friction plus
Stribeck friction plus Coulomb friction plus viscous friction,
the equation of these frictions is expressed as

2
 


 

(12)
f = [fc + (fe fc )e s + B ]sign()

Fig. 1

Consider the n-th order linear time invariant system with


m inputs given by
x(t)

= Ax(t) + Bu(t)

(13)

where A Rnn and B Rnm with 1  m  n. Without


loss of generality, it can be assumed that the input distribution matrix B has full rank. Dene a switching function
s : R Rm as
s(t) = Cx(t)
(14)
where C Rnm is a full rank positive constant matrix
and is set as Hutwitz. Let C be the hyperplane dened by
C = {x Rn : Cx(t) = 0} .

where TL is load torque, B is the damping coecient, and


J is the moment of inertia.
Since the magnetic ux generated from the permanent
magnet rotor is xed with respect to the rotor shaft position, and Lmd and If d are constant for PMSM, d is xed
when substituting id = 0 into (4) and (6). Also, the electromagnetic torque Te is then proportional to iq which is
determined by closed loop control.
Using the eld oriented mechanism, the PMSM drive can
be simplied into
Te = Kt iq

Variable structure control

(15)

Suppose that u(s(t), x(t)) represents a VSC law where the


changes in control strategy depend on the value of the
switching function. It is natural to explore the possibility
of choosing the control action and selecting the switching
strategy so that an ideal sliding motion takes place on the
hyperplane, i.e., there exists a time ts such that
s(t) = Cx(t) = 0,

for all t > ts .

(16)

Suppose that at time t = ts the systems states lie on


the surface s and an ideal sliding motion takes place.
This can be expressed mathematically as Cx(t) = 0 and

from
s(t)

= C x(t)

= 0, for all t > ts . Substituting for x(t)


(13) into the above equation gives
C x(t)

= CAx(t) + CBu(t) = 0,

for all t  ts .

(17)

Suppose that the matrix C is designed so that the square


matrix CB is nonsingular. In practice, this is easily accomplished since B is a full rank and C is a free parameter.
The equivalent control, written as ueq , is dened as the
unique solution to (17)
ueq (t) = (CB)1 CAx(t).

(18)

Equation (18) expresses the control action required to maintain the states on the switching surface. The ideal sliding
motion is then given by substituting (18) into (1). Then,
we have a free motion
x(t)

= (I B(CB)1 C)Ax(t)

Simplied conventional control block diagram for PMSM

(19)

408

International Journal of Automation and Computing 05(4), October 2008

for all t  ts and Cx(ts ) = 0.


Equation (19) shows that the sliding motion is a control
independent of free motion, but dependent on the choice of
sliding surface s although the precise eect is not readily
apparent. An easy way to shed light on the problem is to
rst transform the system into a suitable canonical form.
In this form, the system is decomposed into two connected
subsystems, one acting in R(B) and the other in N (C),
which is referred to as regular form. The second step is to
design the sliding surface s(t) = Cx(t), and then design a
discontinuous control function
 +
u , if S(x) > 0
(20)
u=
u , if S(x) < 0
to move the trajectory of system (13) to surface (14), and to
keep it on the surface thereafter. The following well-known
condition of the existence of sliding mode must be met in
the vicinity of the sliding surface.

S(x)S(x)
< 0.

(21)

A discontinuous control law (20) can be designed in various


formats as (additional input in the form)
u = ueq + u.

(22)

The reaching law approach directly species the dynamics


of the switching function. Let the dynamics of the switching
function be specied by the dierential equation[13, 14] :
s = Qsgn(s) Kf (s)

(23)

where gains Q and K are diagonal matrices with positive


elements, and
sgn(s) = [sgn(s1 ) sgn(sm )]T

(24)

f (s) = [f1 (s) f2 (s)]T .

(25)

The scalar functions fi satisfy the condition


si fi > 0,

for si = 0,

i = 1, , m.

(26)

Then, the sliding condition in (21) is guaranteed.


The main disadvantage of the conventional VSC is the
drastic change of the control signal across the sliding surface, which leads to chattering in the digital implementation. This chattering may damage the control system. The
larger the magnitude of Q is, the more severe the chattering
will be. If Q is too small, the reaching time will be too long.
2) The saturation reaching law is


s(x)
(27)
s = Q sat

where

  s
, for |s| 
s

=
sat
sgn(s), for |s| > .

s = Qsgn(s) Ks.

(28)

4) The power rate reaching law is


s i = ki |si | sgn(s)

(29)

where 0 < < 1, and i = 1, , m.


The reaching law approach not only establishes the reaching condition but also species the dynamic characteristics
of a system during the reaching phase.

Modified adaptive reaching law for


sliding mode controller

As demonstrated in Section 3, the conventional VSC


(Ksgn(s)) is highly robust to uncertainties and load disturbance but causes the chattering problem. Thus, if we
can control the gain K to work when the system is disturbed and minimize the gain in the steady state, then we
can solve the chattering problem. The application of SMC
requires the upper bound of uncertainties and load disturbance. However, the parameter variations of the system are
dicult to measure and the exact value of the external load
disturbance is also dicult to know in advance for practical
applications in industry[11] . The adaptive algorithm is used
here to estimate the rate of change in the upper bound of
uncertainties and load disturbance.
Consider the system
x(t)

= f (x) + Gu(t) + r(t)

(30)

r(t) = f (x) + Gu(t) + d(t)

(31)

where d(t) is the external disturbance (unknown), f (x, t)


is the unknown model error, f (x) is known and f (x) is
unknown but bounded as |f (x)|  F , and the disturbance
d(t) has also the upper bound D. In this way, with conventional VSC as given in (26) and system (30), if
Q +D+F

Equation (23) is called the reaching law. Various choices


of Q and K specify dierent rates for s and yield dierent structures in the reaching law. The following laws are
available[13] .
1) The constant rate reaching law is
s = Qsgn(s).

3) The constant plus proportional rate reaching law

(32)

where is a positive constant, then sliding condition (21)


is guaranteed. Also, if we choose Q in (28) as in (32), then
by applying (28), the sliding equation becomes
ss
 Ks2 |s| .

(33)

Also, the hitting time of the SMC by (26) is longer than


the hitting time of the two-phase SMC by (28).
If we do not know the upper bounds F or D, then r(t)
becomes unknown. Assume that it is bounded as
|r(t)| < E

(34)

where E is an unknown but bounded positive constant. In


[11], the following adaptive algorithm for computing the
rate of change of the upper bound (E) is given by
1

E(t)
= |s(t)CB|

(35)

where > 0, denoting the adaptation gain. Thus, the modied adaptive law becomes

+ Kf (s).
u = ueq + E(t)sgn(s)

(36)

Mohamed Said Sayed Ahmed et al. / Position Control of Synchronous Motor Drive by Modified

Rewrite the reaching law as

s = E(t)sgn(s)
Kf (s).

(37)

The proposed adaptive two-phase sliding mode position


controller is shown in Fig. 2. By using this modied adaptive law, the term Kf (x) is a smoothing term guaranteeing
the fast response in transient and steady state while the

adaptive term E(t)sgn(s)


satises high robustness in disturbed states.

Simulation and results

To design VSC, rst we obtain the regular form, and then


design the sliding surface:
s = Ce + e,

e = d .

The equivalent control is




TL
J
B
) .
C(d ) + (d + +
udeq =
kt
J
J

(38)

(39)

By (22), the conventional variable structure control (CVSC)


is
u = ueq + Q1 sgn(s)
(40)
the two-phase VSC is
u = ueq + Q2 sgn(s) + K2 s.

(41)

Also, by (35) the proposed adaptive two-phase SVSC is

u = ueq + E(t)sgn(s)
+ K3 s.

(42)

We choose the design parameters as C = 100, Q1 = 40,


Q2 = 10, K2 = 0.5, K3 = 5 and according to (29) we put
= 2000. Also, for comparison, we design P-position and
PI-speed controller for the synchronous motor and the simulation parameters as follows: the proportional gain of the
position controller is set at 10, and the proportion and integral gains of the speed controller are set at 5 and 50, respectively. For ction model, fe = 2, fc = 0.95, and s = 0.1.
The parameters of the motor are listed in Table 1.

Fig. 2

409

In the simulation, two cycles of step rotor position commands (step command = 4 rad) are given periodically. The
simulation results of the P-PI controller, the conventional
VSC, two-phase VSC and the proposed adaptive two-phase
VSC start with the nominal case. Thus, disturbance torque
with 2 Nm is given at 0.5 s and the damming coecient is
changed to ve times of the nominal value at 2.5 s. The position response and the associated control eorts are shown
in Fig. 3. Fig. 3 shows that the robust control performance
of the proposed adaptive VSC in both command tracking
and load disturbance is as excellent as conventional VSC,
but unlike the conventional VSC in that there is no chattering in control eort and no ripples in the position of
PMSM. In addition, in the proposed controller we can increase the time response by increasing the gain K with no
eect on chattering phenomena and robust performance.
Figs. 4 and 5 show the position response of the proposed
controller to sinusoidal tracking and square tracking, respectively. Fig. 6 compares the proposed controller with
the conventional PID controller when changing the nominal case to disturbed case at 1.5 s. (the load torque TL
The symbols
changed from 2 N to 3 N, j = 10
j, B = 10B).

j and B represent the corresponding system parameters in


the nominal condition. The parameters for PID are given
as KP = 200, KI = 10, and KD = 10. The parameters for
the proposed controller are mentioned previously. For the
proposed controller, when the disturbed state is applied,
to overcome
the controller will change the value of Q(E)
this disturbances, so the disturbances are proportional to
However, in conventional PID or CVSC (Q1 , Q2 , K2 ),
E.
the gains are xed, not depending on the value of disturbances. We note that the modied controller has faster
response than the conventional VSC and has no ripples in
the resulting position response. The modied controller is
free of chattering in the steady state and undisturbed state.
When the system is subjected to parameter variations and
external disturbances, the conventional VSC is very insensitive to parameter variations and external disturbances,
while the modied controller is similar to the conventional
VSC. For the modied controller, we can increase the robustness by controlling , if we set at a very large value,
the modied controller is reduced to one-phase VSC.

PMSM with adaptive two-phase sliding mode position controller

410

International Journal of Automation and Computing 05(4), October 2008

(a) Position response of P-position and PI-speed controller

(c) Position response of conventional VSC controller

(e) Position response of two-phase VSC controller

(g) Position response of adaptive two-phase VSC controller

Fig. 3

(b) Control input of PID controller

(d) Control input of conventional VSC controller

(f) Control input of two-phase VSC controller

(h) Control input of adaptive two-phase VSC controller

Position responses of the conventional and proposed controllers (nominal case with TL given at 0.5 s and change of B at 2.5 s )

Mohamed Said Sayed Ahmed et al. / Position Control of Synchronous Motor Drive by Modified
Table 1

Torque constant
0.6732 Nm/A

411

Specications of PMSM

Stator

Flux

Damping

Rated

Stator

Moment of

resistance

linkage

coecient

torque

inductance

inertia

1.47

0.2244 Wb

0.093 103 Nms/rad

2.5 Nm

5.33 mH

1.32 103 Nms2

Conclusions

A design concept for the theory of VSC is presented. In


this paper we introduced a new adaptive two-phase VSC for
position control of a synchronous motor which has fast response and highly robust against uncertainties and external
disturbances. The proposed controller uses two-phase SMC
where the 1st phase is a smoothing phase which mainly controls the system in steady states and disturbed states; while
the 2nd phase is an adaptive phase which is used mainly in
the case of disturbed states. Using our modied reaching
law, the controller can be used in practice. In our future
work, the two-phase of the reaching law will be tuned using
a fuzzy controller.
Fig. 4 Position response of the proposed controller to sinusoidal
tracking

References
[1] J. J. E.Slotine. Sliding Controller Design for Nonlinear Systems. International Journal of Control, vol. 40, no. 2, pp.
421434, 1984.
[2] J. J. Slotine, S. S. Sastry. Tracking Control of Nonlinear
Systems Using Sliding Surface with Application to Robot
Manipulators. International Journal of Control, vol. 38, no.
2, pp. 465492, 1983.
[3] H. Lee, D. Nam, C. H. Park. A Sliding Mode Controller
Using Neural Networks for Robot Manipulator. In Proceedings of European Symposium on Artificial Neural Networks,
Bruges, Belgium, pp. 193198, 2004.

Fig. 5
Position response of the proposed controller to square
tracking

[4] S. Elangovan, P. Y. Woo. Adaptive Fuzzy Sliding Control


for a Three-link Passive Robotic Manipulator. Robotica,
vol. 23, no. 5, pp. 635644, 2005.
[5] K. K. Shyu, Y. W. Tsai, C. F. Yung. A Modied Variable
Structure Controller. Automatica, vol. 28, no. 6, pp. 1209
1213, 1992.
[6] W. Gao, J. C. Hung. Variable Structure Control of Nonlinear Systems: A New Approach. IEEE Transations on
Industrial Electronics, vol. 40, no. 1, pp. 4555, 1993.
[7] W. J. Wang, J. L. Lee. Hitting Time Reduction and Chattering Attenuation in Multi-input Variable Structure Systems. Control Theory and Advanced Technology, vol. 9, no.
2, pp. 491500, 1993.
[8] D. Q. Zhang, S. K. Panda. Chattering Free and Fast Response Sliding Mode Controller. IEE Proceedings of Control
Theory and Applications, vol. 146, no. 2, pp. 171177, 1999.

Fig. 6
Position response of the proposed controller and PID
controller when the nominal case is changed to disturbed case at
1.5 s

[9] A. Levant. Higher-order Sliding Modes, Dierentiation and


Output-feedback Control. International Journal of Control,
vol. 76, no. 9, pp. 924941, 2003.

412

International Journal of Automation and Computing 05(4), October 2008

[10] A. J. Koshkouei, K. J. Burnham, A. S. I. Zinober. Dynamic


Sliding Mode Control Design. IEE Proceedings of Control
Theory and Applications, vol. 152, no. 4, pp. 392396, 2005.
[11] F. J. Lin, S. L. Chiu, K. K. Shyu. Novel Sliding Mode Controller for Synchronous Motor Drive. IEEE Transactions on
Aerospace and Electronic Systems, vol. 34, no. 2, pp. 532
542, 1998.
[12] P. Pillay, R. Krishnan. Modeling, Simulation and Analysis of Permanent Magnet Motor Drives, Part I: The
Permanent-magnet Synchronous Motor Drive. IEEE Transactions on Industry Applications, vol. 25, no. 2, pp. 274
279, 1989.
[13] J. Y. Hung, W. Gao, J. Hung. Variable Structure Control:
A Survey. IEEE Transactions on Industrial Electronics, vol.
40, no. 1, pp. 222, 1993.
[14] Y. B. Shtessel. Sliding Mode Control of the Space Nuclear
Reactor System. IEEE Transactions on Aerospace and Electronic Systems, vol. 34, no. 2, pp. 579589, 1998.
Mohamed Said Sayed Ahmed graduated from the Military Technical College (MTC), Egypt, in 1995. He received
his M. Sc. degree from Cairo University,
EGYPT, in 2001. He is currently a Ph. D.
candidate in Automation Science and Electrical Engineering School at Beijing University of Aeronautics and Astronautics
(BUAA), PRC.

His research interest includes nonlinear control, especially sliding mode control, fuzzy control, neural control, electrical drives,
and ight control.

Ping Zhang
graduated from Beijing
University of Aeronautics and Astronautics
(BUAA), PRC, in 1976. She received her
M. Sc. degree from BUAA in 1982 and
Ph. D. degree from BUAA in 2001. She is
currently a professor in Department of Automatic Control, Automation Science and
Electrical Engineering School of BUAA.
Her research interest includes ight control and simulation,
fault tolerant control, faults detection and diagnosis, machine
vision, gure and image pattern identication, and virtual prototype.

Yun-Jie Wu
graduated from Harbin
Engineering University, PRC, in 1991. She
received her M. Sc. degree from Harbin Engineering University in 1994 and Ph. D. degree from Beijing University of Aeronautics
and Astronautics (BUAA), PRC, in 2002.
She is currently a professor in Department
of Automatic Control, Automation Science
and Electrical Engineering School of BUAA.
Her research interests include motion control, process control,
and intelligent control.

Anda mungkin juga menyukai