Discrete-Time Sliding Mode Control of Permanent Magnet

Linear Synchronous Motor in High-Performance Motion

with Large Parameter Uncertainty
M. Jalili-Kharaajoo and R. Fazaie
Control & Intelligent processing Center of Excellence,
ECE Department, University of Tehran,
P.O. Box 14395/515, Tehran, IRAN
Tel: (+98-21) 8020403, Fax: (+98-21) 8778690
E-mail: [email protected] and [email protected]
Abstract: In order to meet the demands of high acceleration/deceleration and high-precision motion
profiles, many motion control systems began to use direct-drive linear motor as the prime motion
actuator. This arrangement has the advantage of providing high-performance motions with the advantage
of providing high-performance motions with reduced mechanical components, but its major drawback is
the effect of load variation on the overall system control. Unlike conventional ball-screw drive, a linear
direct-drive system eliminates the mechanical couplings, rotary-to-linear translators, and reduction gears.
Under this arrangement, any change or disturbance in the load will be directly reflected back to the motor
and the control system. This will cause large deterioration in the motion profile. In this paper, a robust
controller based on Digital Sliding Mode Control (DSMC) is proposed to overcome the load uncertainty
problem. In the investigation, a permanent magnet linear synchronous motor (PMLSM) with large
parametric uncertainty is chosen as the target study. First, the state space equations of the motor are
established. Then the DSMC theory is applied to design a robust controller. The simulation results show
that the system can achieve robust performance under large load variations. Thus, the proposed method is
an effective mean of combating load variations and load disturbances in high-performance direct-drive
systems.
Keywords: Permanent magnet linear synchronous motor, discrete-time sliding mode control, motion
control, parameter uncertainty, robust control.
1. Introduction
With the recent advances of computer technologies, control
theories and material technologies, more and more linear
motors are employed in application ranging from mass
transportation to factory automation [1,2,3]. Among these
applications, permanent magnet linear synchronous motors
(PMLSM) are frequently used duo to their simple structure,
ease of manufacture, high power density, large torque to
inertia ratio, high efficiency and simple control. The wide
variety of applications and increasing use of PMLSM motor
drives make it necessary to come up with fast and reliable
motor drive control system design approaches.
The continuous-time variable structure controller is known
to be robust to parameter uncertainty and external
disturbance because of the sliding motion on a predefined
hyperplane. The most distinguishing property of variable
structure control systems (VSCS) is that the closed loop
system is completely insensitive to system uncertainties and
external disturbances [4]. VSCS, as the name suggests, are
a class of systems whereby the "control law" is deliberately
changed during the control process according to some
defined rules, which depends on the state of the system [5].
Continuous-time SMC has shown superb characteristics in
terms of insensitivity to large parameter variations and
complete rejection of exogenous disturbances [6].
However, when digital computers are involved in the
control implementation, chattering commonly arises due to
the limited sampling rate and discretization errors. To solve
this problem, the Discrete-time Sliding Mode (DSM)
concept has been introduced and much research on DSM
control (DSMC) has also been presented in recent years [7-
11]. In sampled-data systems, the complete disturbance
rejection cannot be achieved since the zero-order-hold does
not take place in the disturbance channel. In [12] it is
shown that DSMC for nominal plants can be modified to
compensate matched disturbances to O(T
2
) accuracy (T is
the Sampling time).
In this paper, we study the case of PMLSM used as
transportation system in factory automation. For this
purpose, the mass of the moving part is varies frequently
under no-load and full-load conditions. The friction of the
motion system also varies frequently and significantly.
Therefore, it is necessary to design a robust controller that
is insensitive to the variation of the mass and friction, to
SICE Annual Conference in Fukui, August 4-6, 2003
Fukui University, Japan
PR0001/03/0000-0760 400 2003 SICE -760-
ensure that the system is stable system under arbitrary
loads. Obviously, it is difficult to meet above specifications
using conventional PID controllers. PID controller cannot
make the system stable under certain conditions. If applying
linear system theory can solve the control problem, then the
accurate mathematical model of the system is required,
however, the variation of mass and friction causes large
parameter uncertainty in the system. One of the control
strategies that seems to be efficient method in this situation
is the use of sliding mode control. Although, continuous-
time SMC has shown superb characteristics in terms of
insensitivity to large parameter variations, it can produce
significant chattering in the system. To solve this problem,
the discrete-time sliding mode control strategy is used. The
use of discrete-time sliding mode control can decrease
chattering in addition to guarantee the robust performance
of the closed-loop system in response to parameter
uncertainties. In this approach the exact model of the
system is not required and some uncertainties allowed. The
paper analyses the system dynamic characteristics, deduced
the state space model of the system, and designs a discrete-
time sliding mode controller for the PMLSM system.
Firstly, the d-q dynamic model for the PMLSM is studied.
The d-q coordinate system is a "rotating" reference frame
that moves at synchronous speed. Then, the motion
equation of the system is obtained according to Newton's
mechanics law that leads to the state space model of the
system. All simulations are carried out by MATLAB. From
simulations, the closed-loop system with the proposed
controller exhibits excellent performance in the presence of
uncertainty in the mechanical parameters of the motor. In
addition, the close-loop system with the proposed control
action has robust performance in response to load torque
variations.
2. Dynamic model of the system
Firstly, the d-q dynamic model for the PMLSM is studied.
The d-q coordinate system is a "rotating" reference frame
that moves at synchronous speed. The flux linkage
equations are as follow
(
(
(

(
(
(

=
(
(
(
(

f
q
d
f
q
f d
f
q
d
i
i
i
M
L
M L
0 0
0 0
0

(1)
where
f q d
, , are the flux linkage of direct axis,
quadrature axis and permanent magnet, respectively.
f q d
M L L , , and
f q d
i i i , , are inductance and current of d-,
q-axis and equivalent permanent magnet, respectively. For
surface PMs, we have
q d
L L = [2].
f f
M , and
f
i are
all constant.
The voltage equations of the PMLSM are shown in (2)
(

+
(

+
+
=
(

f f q
d
q d
q d
q
d
i M i
i
pL R L
L pL R
u
u
e e
e 0
(2)
where
d
u ,
q
u are d-, q-axis armature voltage, R is phase
resistance, is equivalent "electrical angular velocity",
t vt e / = , v is the velocity of moving part, is the pole
pitch, p is derivative operator, dt d p / = .
The electromagnetic thrust is
] ) ( )[ / ( 5 . 1
) )( / ( 5 . 1 /
q d q d q f n
d q q d n elm m
i i L L i p
i i p P F
+ =
= =
t t
t t v
(3)
where
m
F is the output electromagnet thrust,
elm
P is the
electromagnet power,
n
p is the number of pole-pair(s). In
this paper, 1 =
n
p .
Lastly, the motion equation of the system is obtained
according to Newton's mechanics law as
m v R f F dt dv
v m
/ ) ( / = (4)
where m is the mass of moving part including the load, f is
the total friction,
v
R is the damper coefficient associated
with velocity.
When 0 =
d
i control scheme is applied, the d-axis flux
linkage is equal to the permanent magnet flux linkage. The
dynamic model of the system can be simplified considering
the velocity v, q-axis current
q
i and total friction
coefficient as state variables; we obtain the state equation
and output equation of the system as follow
Bu Ax x + =
.
, cx y = . (5)
where
,
(
(
(

q
i
v
x
(
(
(
(
(
(
(

=
0 0 0
0
1 5 . 1
q q
f
f
v
L
R
L
m m
R
A

t
t

t
t
,
(
(
(
(

=
0
1
0
q
L
B ,
q
u u = , | | 0 0 1 = c
where the mass of the moving part when under rating load
is up to 10 times of that when under no load.
3. DSMC design
The VSS theory was originally developed from a
continuous time perspective. It has been realized that
directly applying the continuous-time SMC algorithms to
discrete-time systems will confront some unconquerable
problems, such as the limited sampling frequency,
sample/hold effects and discretization errors. Since the
switching frequency in sampled-data systems cannot
-761-
exceed the sampling frequency, a discontinuous control
does not enable generation of motion in an arbitrary
manifold in discrete-time systems. This leads to chattering
along the designed sliding surface, or even instability in
case of a too large switching gain. Consider the dynamic
system as follows
Bd t Bu t Ax t x + + = ) ( ) ( ) ( (6)
where d consists of the parameter uncertainty and
disturbances. The discrete-time representation of the system
can be obtained by applying a zero-order-hold sampling
with sampling period T to the continuous-time process. This
gives
k k k k
d u x x I + I + u =
+1
(7)
where u, I are obtained from the following equations over
one sampling time interval, t e[kT,(k+1)T|

At
e = u ,
)
= I
T A
B d e
0
t
t
(8)
In general, holding the continuous-time matching condition
does not necessarily mean that the same condition also
holds in a discrete-time system since the zero order-hold
does not take place in the disturbance channels. Hence, the
disturbance may not be rejected completely even if
discrete-time sliding mode occurs. However, the error
introduced due to the fact that the disturbance does not
satisfy the matching condition is O (T
2
). In other words, this
discrete-time model is an O (T
2
) approximation of the exact
model described by the same u,I matrices. From an
engineering design perspective, the O(T
2
) approximate
models are usually adequate since the inter-sampling
behavior of the continuous-time process is also close within
O (T
2
) to the state values at the sampling instants [5].
The DSMC is supposed to steer the state of the system (7)
towards the sliding surface S and then maintain it on S at
each sampling instant, such that
0 = A =
k k
x s
By solving s
k+1
= 0, the equivalent control for this
perturbed system is obtained as
k k
eq
k
d x u Au AI =
1
) ( (9)
The above control action is not realizable since
computation of u
k
eq
requires complete information of the
disturbance d
k
. To solve this problem, one-step delayed
disturbance estimation was used in [8,9,12], i.e., to compute
the disturbance in the previous sampling instant as
1 1 1
I u = I
k k k k
u x x d (10)
Then d
k-1
is used to approximate the current disturbance d
k
,
such that the equivalent control (9) is approximated by
1
1
) (

Au AI =
k k
eq
k
d x u (11)
The very promising feature of this method is that it does not
require the knowledge of the disturbance bounds, which
makes the control design very simple. Applying the control
(9) to (7), resulting in the closed-loop dynamics, can
immediately see the effectiveness of the controller
) ( ) ) ( (
1
1
1

+
I + u A AI I =
k k k k
d d x I x (12)
where by denoting Ad
k
=d
k
-d
k-1
, which IAd
k
is the control
error, or by pre-multiplying to both side of (12), the error
can actually be expressed by
k k
d s AIA =
+1
(13)
We can see that if d
k
is a constant disturbance,
1 + k
s equals
zero and does not depend on the magnitude of d
k
.. It has
also been proven that if the disturbance satisfies the
boundedness and smoothness conditions, the resulting
control law is able to constrain the system to stay on an
O (T
2
) boundary layer of the sliding surface [12].
It is known that when the state is not on S, the equivalent
control (9) tries to drive the state onto the sliding surface in
one sampling period. However, if the initial state is far
away from S, since the magnitude of the control (9) may
become infinitely large as T 0, this requirement is far
beyond the actuators limitation. Two methods have been
proposed for practical implementation of the control law.
One method that has been proven in [10] takes into account
the actuator saturation value u
m
, by the control law

>
s
=
m
eq
k
eq
k
eq
k
m
m
eq
k
eq
k
k
u u if
u
u
u
u u if u
u (14)
The system can be forced into the vicinity of discrete-time
sliding mode in a finite number of steps.
4. Simulation results
The motor parameters used in the simulation are listed as
follows:
Phase resistance R: 4.5 ohm;
Phase synchronous inductance L
q
: 3 mH;
Permanent magnet flux linkage
f
: 0.3438 V.s/rad;
Pole pitch of the permanent magnets t : 0.03048 m;
Number of pole pairs P
n
: 1;
Total mass of moving part excluding load m: 1.635 kg;
Viscous damping coefficient R
v
: 0.1 N.s/m;
Total friction coefficient : 1 N/kg.
In the simulations, sampling time is chosen as T=0.001sec.
Fig. 1 shows the simulation result of velocity step response
with mass variation range from 0 to 100 percent of nominal
load. The closed-loop system has excellent performance
-762-
when the mass is 5 times of system mass, the overshoot is
inconsiderable and the settling time is short. When the mass
decreases, the overshoot goes down but the settling time
increases. On the contrary, when the mass increases, the
overshoot increases and the settling time also increases.
However, the closed-loop system is stable no matter how it
is under any load. So, the controller is robust even if the
controlled plant has large parametric uncertainty.
The actual velocity step responses under no-load and full-
load are depicted in Fig. 2. In this simulation, two worst
cases have been tested. One is that the motor starts without
load, and then full load is suddenly exerted at 1.2sec. The
other case is that the motor starts with full load, and then
the load is suddenly emptied at 1.2sec. Evidently, the
closed-loop system is stable no matter how much load is
carried. However, if this plant is controlled by a PID
controller, the controller coefficients must be modified
frequently with the variation of the total mass in real time.
Additionally, the whole system will be unstable when it is
under full load. Even if it is stable, the step response is
worse with large overshoot and long settling time.
Fig. 1. Simulated velocity step responses with different
masses.
Fig. 2. Simulated velocity step responses under no-load
and full-load conditions.
5. Conclusion
In this paper, discrete-time sliding mode control strategy
was applied to speed control of permanent magnet linear
synchronous motors. Due to sliding nature of DSMC, this
control algorithm has robust performance in response to
external disturbance and parameter uncertainty.
Additionally, DSMC is a chattering-free control scheme.
Simulation results showed that the proposed control action
has robust performance in response to load variations and
system parameters uncertainty.
6. Reference
[1] S. W. Tam and N.C. Cheung, An all-digital high-
performance drive system for linear permanent-magnet
synchronous motor, The 14
th
Chinese Power Supplies
Society Annual General Meeting, Beijing, China, 24-27 Sep
2001.
[2] J.F. Gieras and Z.J. Piech, Linear synchronous motors:
Transportation and automation systems, CRC Press, 2000.
[3] N.C. Cheung, A robust and low cost linear motion
system for precision manufacturing automation, The 35
th
IEEE Industry Application Society Annual Meeting,
IAS'2000, Roma, Italy, Vol.1, pp40-45, 8-12 Oct 2000.
[4] Edwards, C. and Spurgeon, S. K. Sliding mode
control. Taylor & Francis Ltd, 1998.
[5] Young, D. K., Utkin,V. I. and zgner. A control
engineers guide to sliding mode control. IEEE Transaction
on Control System Technology 7(3), May, 1999.
[6] Bartolini, G. and Pydynowski, P. An improved,
chattering free, VSC: scheme for uncertain dynamics
systems. IEEE Transaction on Automatic Control, 41(8),
August, 1996.
[7] Pan, Y. and Furuta, K. Discrete-time VSS controller
design, Int. J. Robust Nonlinear Control, 7, 373-386, 1997.
[8] Su W.C., Drakunov S.V., zgner, U., and Young,
K.D. Sliding Mode with Chattering Reduction in Sampled
Data Systems. Proc. 32
nd
Conference on Decision and
control, 1993.
[9] Su, W.C., Drakunov, S.V., and zgner.
Implementation of Variable Structure Control For Sampled
Data Systems. ISBN 3-540-76067-9 Springer- Verlag
Berlin, Heideberg, New York, 1996.
[10] Utkin, V.I. Sliding mode control in discrete-time and
difference systems. Variable Structure and Lyapunov
Control. Alan S.I. Zinober (Ed.) Springer- Verlag London
Limited. 87-107, 1994.
[11] Jalili-Kharaajoo, M., Ebrahimirad, H., Yazdanpanah,
M.J. and Labibi, B., Digital sliding mode control applied to
position control of induction motors, to appear in IFAC
ROCOND2003, Italy, June, 2003.
[12] Su, W.C., Drakunov, S.V. and zgner. U. An O (T
2
)
boundary layer in sliding mode for sampled data systems,
IEEE Trans. Automatic Control, 45(3), 482-485, 2000.
-763-