International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________

Applying Digital MIMO Dynamic Sliding Surface Control to Idle Speed Control
of Spark Ignition Engine
Chin-Wen Chuang

Ching-Sheng Chi

Department of Electrical Engineering

I-Shou University
Kaohsiung, TAIWAN, R.O.C.
cwchuang@isu.edu.tw

Department of Electrical Engineering

I-Shou University
Kaohsiung, TAIWAN, R.O.C.
michel@isu.edu.tw

AbstractThe good idle speed control of spark ignition engine makes the customers comfortable and drives smooth. To derive its
controller, a control-oriented object based model of spark ignition engine is discussed in this paper. To achieve accuracy tracking
and reduce the control effort, a digital MIMO dynamic sliding surface control rule is proposed in this paper. The sliding mode
controller will drive the system to tracking the desired response. The existence condition of the discrete sliding mode is expanded
to two conditions. The first one is used to keep the sliding motion toward to the sliding surface. The next one is to keep the sliding
motion convergent. The dynamic sliding surface will decrease the approaching time. At the same time, the output magnitude of
the switching part will be reduced efficiently. Finally, the simulation results will demonstrate the performance of the proposed
control rule. The different working conditions will show the robustness of the proposed controller for parameter variation and the
external load.
Keywords- dynamic sliding surface control, idle speed, ignition engine.

__________________________________________________*****_________________________________________________
I.

INTRODUCTION

The engine is wide applied in industrial application [1-11].

The different kinds of engine, such as diesel engine, spark
ignition engine and etc., have been developed in recent
researches [1-6]. The spark ignition engine is especially applied
to the personal cars. The stability of idle speed of spark ignition
engine will affect the user's comfort. Therefore, the major
objective is the idle speed control of a spark ignition engine in
this paper. A linear model of the Fiat Dedra engine [7] is used
to study the idle speed control problem. The engine dynamics
based on the intake manifold process, combustion and
rotational dynamics compose a multi-input multi-output
(MIMO) system. A time delay caused by the transports of air
and fuel in combustion process is also present and affects this
system. At the same time, the parameters of the linear model
measured on different operating conditions are so different that
there exist the uncertainty problems in design of such engine
systems. Therefore, the controller of the idle speed must be
able to overcome the parameter deviations and against the
external disturbances.
The variable structure control (VSC) [12] has been applied
to many industrial fields. The advantages of the order-reducing
and against to external disturbance cause the proper robustness
of the VSC. However, in order to lock on the sliding surface,
the hard switching method causes the chattering phenomena.
The high frequency switching does not make the VSC to be
practical [13-14].To smooth the chattering, some soft switching
function is taken to replace the sign function. At the same time,
the additional integral controller is used to eliminate the steady
state error. Mostly variable structure controller is divided into
the equivalent controller and the switching controller. The first
controller is working under the ideal condition. The second
controller is used to force the sliding motion toward the sliding
surface and make the stability. Hence the integral gain and the
switching gains are selected to satisfy the existence condition,
that is

t t 0

Change the differential equation to difference equation in

order to apply to the discrete system. When applying the
condition to discrete systems, the existence condition becomes
to

k k 1 k 0

However, referring to the reference [15], the system is

controlled by the controller only in each sampling time. The
controller cannot modify the response during the sampling
interval. It may happen that the condition (2) is not only
satisfied but also the sliding motion is divergent, shown in fig
1.
Phase Plane

(k+2)
(k)
(k+3) (k+1)

Figure 1. Discrete sliding mode

The condition (2) only makes the sliding motion toward to

the sliding surface. However, it cannot guarantee the sliding
mode convergent to this surface. The condition (2) is only the
necessary condition not the sufficient condition in discrete
systems. To make up the drawback, the reference [15]
introduces one additional restriction, that is

4118
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________

k 1 k

Phase Plane

Combining equation (2) and (3) can make sure the sliding
motion convergent. However, the sliding surface is changed to
sliding region shown in fig 2.

0
(k+1)
(k)
(k+3)
(k+2)
(k+4)

(k+5)

Phase Plane

(k+3)

Figure 4. Discrete sliding mode with dynamic sliding surface

(k+4)
(k+5)

(k+2)

This method uses the first order of difference equation of

sliding variable, that is

(k)

(k+1)

Figure 2. Discrete sliding mode with sliding region

The choice of the switching gain becomes three states. This

change causes some difficulty in implementation of hardware.
To maintain the binary choice, one restriction of different
viewpoint is introduction in reference [16], that is

k 1 p k , 0 x0

k 1 k

where is a small positive constant. The varying of each

step of sliding motion is restricted. Then, condition (2) makes
the sliding mode toward to the surface. The condition (4)
makes the sliding motion oscillated on this surface within a
small range shown in fig 3.
Phase Plane

The initial of systems states sets the initial value of the

difference equation. Obviously, at the beginning the systems
dynamic is positioned on the dynamic sliding surface. Hence,
the reaching can be reduced efficiently. Only small switching
gain in needed to eliminate the parameter variation and external
disturbance. Therefore, the chattering phenomena can be
eliminated.
The novel digital MIMO dynamic sliding surface Control is
proposed in this manuscript to apply to the idle speed control of
spark ignition engine. To make sure the existence of sliding
motion, the continuous condition is expanded to two conditions
[13-14]. The existing conditions of the digital sliding mode are

k 1 k k 0
i

and i k 1 i k i , where i k
2

means each sliding surface. The dynamic sliding surfaces are

defined as S m1 k 1 Pmm S m1 k . The detailed design
procedures will be discussed in this manuscript. Finally, the
simulation results will demonstrate the potential of the
proposed approach even the external disturbances existed.
II.

THE LINEAR MODEL OF THE SPARK IGNITION ENGINE

The block diagram of the linear model of the Fiat Dedra

engine is shown in fig 5. It is consisted of three main physical
phenomena inner the engine. They are intake manifold
dynamics, combustion and rotational dynamics.

(k+2)
(k)
(k+3)
(k+1)

To reduce the time of out of control when the motion is not

fixed on the sliding surface, the high switching gain is usually
chosen to speed up the reaching time. However, in discrete
systems the controller only modifies the control signal at each
sampling time. High switching gains can speed up the reaching
time, but the chattering often be enlarged. To eliminate the
chattering and decrease the reaching time, this paper introduces
the dynamic sliding surface control (DVSSC) rule shown in fig
4.

Combustion

Intake Manifold Dynamics

Figure 3. New discrete sliding mode

1
s

Rotational Dynamics

GP

GF

GA

e ds

TL

1
Js

Bm

KP

+
KN1

KN 2

Figure 5. Block diagram of the linear model of the engine

In the intake manifold dynamics, one can assume that the

temperature is approximately constant. Therefore, one can treat
the modeling development of the manifold as the air pump. The
4119
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________
relationship between the rate of change of the manifold
pressure Pt and the duty cycle of the throttle valve Dt is
P t K Dt K N 1 N t K p Pt

where K is the constant gain, K N 1 is the pumping

feedback gain and K p is the gas gain.
The combustion process generating the required torque is
depended on the air, fuel F t , engine speed N t and spark
advance position At . Those elements don't have immediately
influence on the combustion process except the spark advance
position. A time delay called induction-to-power (IP) stroke
lag, d , is produced by the transport delay. The amplitude of
the time delay depends on the number of the cylinders and the
valve in applied engine [1]. The relationship of the combustion
process can be described as

R Y
Z
p1
p1
p1

Te t G p Pt GF F t K N 2 N t delay d GA At

where

z1
r1
r
z
2
2
Z p1 , and R p1 .

rp
z p
Then, by forward difference to replace the analog
differential operation, the system model will be re-written as
TC p n Z p1 k 0 pm
Z p1 k 1 I p p

X k 1 0

U m1 k
I
n1
n p n n TA n n Xn1 k TB n m

TI p p
0 pn

R p1 k
Fn1 k
TI n n
0n p

where G p , G F , K N 2 and G A are the relative gains. Finally,

the rotational dynamic is given by the Newton's second law as

JN t Te t TL t Bm N t

Let the sliding surface is chosen to be

THE DYNAMIC DIGITAL MIMO SLIDING SURFACE

CONTROL

If the MIMO system is defined as n states and p outputs,

that is

A X B U F
X
n1
nn
n1
nm
m1
n1

Yp1 C pn X n1

where

x1
a11 a12 a1n
x
a
a22 a2 n
,
,
X n1 2
A nn 21

xn
an1 an 2 ann
b11 b12 b1m
u1
f1
b

f
b22 b2 m
u2
21

, U m1
, Fn1 2 ,
B nm

bn1 bn 2 bnm
u m
fn
y1
a11 a12 a1n
y
a
a22 a 2 n
2
Yp1 , and A nn 21

y p
an1 a n 2 ann
Let the output error integral controller is included in this
system, that is

Yp1 k C pn X n1 k

S m1 k 0m p Yp1 k K p p Z p1 k mn X n1 k

where J is the engine inertia, TL t is the external torque

load and Bm is viscous friction constant.
III.

0m p K p p

Z k
0m p C pn mn p1
X n1 k

where 0m p and mn are switching gain matrix

the integral gain matrix. They are defined as
110 120
1
0

21 220
2
0

,
m p
S m1

0

0
n
m1 m 2

and K p p is

10p

20 p

0
mp

11 12 1n

22 2 n
, K p p diagK1 , K1 , , K p ,
mn 21

m1 m 2 mn
and Pmm diagp1 , p1 , , pm
At the same time, let the dynamic sliding surface be an one
ordered differential function as

S m1 k 1 PmmS m1 k

If the sliding motion is locked on the dynamic sliding

surface, then substitute eqs. (9~10) into eq. (11), one has

4120
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________

S m1 k 1 Pm m S m1 k

T 0m p C p n m n A n n

I m m Pm m 0m p C p n m n

0
T m p K p p C p n

T
T

F k
K Z k

0
m p

C p n m n B n m U m1 k

0
m p

C p n m n

R k

I m m Pm m 0m p

T 0m p K p p

X n1 k

p p

p 1

Dmm diagd1 , d 2 , , d m

Then one can solve the coupling effect of U m1 k in

S m1 k 1 S m1 k 1 .
Let the systems parameters exist the nominal values and
the derivate values, that is A nn A 0nn A nn ,

B nm B B nm , Dmm D Dmm . Let the control

function be transferred to diagonal form, that is
~
0m p C pn mn B 0nm B nm U m1 k D 0mm D mm U m1 k
0
mm

where
a11 a12
a
a22
21
A nn

an1 an 2

~s
0
U m1 k m p k Yp1 k K p p Z p1 k mn k X n1 k m1 k

p 1

Let the switching control is defined as eq (9), that is

where

~
U m1 k Dm1m 0m p C pn mn B nm U m1 k

0
nm

If there exists one diagnostic matrix such that

D0m m I m m Pm m 0m p K p p Z p 1 k T D0m m 0m p K p p R p 1 k

T0m p K p pC p n X n1 k

n 1

~
1
U eqm1 k D0m m T 0m pC p n m n A n n I m m Pm m 0m pC p n m n

where the switching gain matrix is

110 k 120 k 10p k
0

21 k 220 k 20p k
,
0m p k

0
0
k
mp
m1 k m 2 k

11 k 12 k 1n k
1 k
k k k

22
2n
, k 2 k .
mn k 21
m1

mn k
m1 k m 2 k
m k
Then, the difference of the sliding surface will be
S m1 k 1 S m1 k

I m m D m m D 0m m 1 I m m Pm m S 0m p

Y p1 k K p p Z p1 k
0
T m p k

T S 0m p C p n S m n A n n I m m D m m D 0m m I m m Pm m S m n

1
TD m m D 0m m S 0m p C p n S mn S 0m p K p p C p n
X n1 k

T m n k
0
0
TD m m D m m S m p K p p R p1 k

0
T S m p C p n S m n Fn1 k Tm1 k

a1n
a2 n
,

ann

b11 b12 b1m

b
b22 b2 m
, and
B nm 21

bn1 bn 2 bnm
Dmm diagd1 , d 2 , , d m .
Let the control function can be divided into the equivalent
controller and the switching controller, that is
~
~
~
U m1 k U eqm1 k U ms 1 k . The equivalent controller is used
to guide the sliding motion along the sliding surface without
the parameter variation and external disturbance. The switching
controller is used to force the motion to move toward the
sliding surface. According to eq. (12), the equivalent control
function can be found as

To satisfy i k 1 i k i k 0 for i 1,, m , one

has

d
ij0 k ij0 sup1 0 i 1 p i ij0 Td i
di

if y j k k j z j k i k 0
ij0 k

d i
0
0
0
ij k ij inf 1 d 0 1 p i ij Td i

if y k k z k k 0
j
j j
i

4121
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________
n

T il c lq iq a qj
q 1 l 1

ij k ij sup
1 d i 1 p

i
ij

d i0

p
d p

T 0 i il0 c lj ij il0 k l c lj

d i l 1
l 1

if
x
k

j
i
ij k
n

il c lq iq a qj

q 1 l 1

ij0 k ij inf

d
1
i
1 p i ij

0

d
i

p
p
d i

il0 k l c lj
il lj
ij

0
d i l 1
l 1

if x k k 0
j
i

d i p 0

T 0 ij k j r j k

d i j 1

i k i sup
p
n

T il c lj ij f j k

j 1 l 1

if i k 0

i k

d i p 0

T
ij k j r j k

d i j 1

i k i inf
p
n

0
T

f
k

il
lj
ij
j

j 1 l 1

if i k 0

To satisfy i k 1 i k
p

let

0
ij

j 1

j 1

ij

i
2

i
2

Td i

d
1 0 i 1 p i ij0 Td i ij0 k
di
i k 1 i k


j 1

y
k

k
z
k
j j
j

j 1

Td i

q 1

0
il

l 1

d i
d i0

0
ij

j 1

q 1

l 1

1 p i ij

k l c lj Td i ij k x j k

0
il

j 1

lj

ij

l 1

f j k Td i i k

1 p i ij0 Td i ij0 k y j k k j z j k

d i
d i0

0
il

T
n

k j r j k T

j 1

c lq iq a qj 1 0 i
di

d i p
il0 c lj ij
d i0 l 1

j 1

0
il

l 1

c lq iq a qj 1 0 i
di

i 1p ,, m,

d i p
il0 c lj ij
d i0 l 1

0
il

1 p i ij

j 1,, n

k l c lj Td i ij k x j k

l 1

n
Td i
d i p
p

ij0 k j r j k T il0 c lj ij f j k Td i i k
T d0
j 1
j 1 l 1

p
n

i ij0 ij i
2
j 1
j 1

Td i

To solve the above equation, one can has the limits of the
switching gains as L0 0 M 0 , i 1,, m, j 1,, n and

ij
ij
ij

for i 1,, m , one can

for i 1,, m . Then

Lij ij M ij , i 1,, m,
i 1,, m , where

d
L0ij k sup1 0 i
di

j 1,, n and Li i M i ,

i 1,, m

1 p i ij0

Td
i

sup ij0 Td i y j k k j z j k

M ij0 k inf 1 0 i 1 p i ij0 Td i

d

0
sup ij Td i y j k k j z j k

d
p

Lij sup T il0 c lq iq a qj 1 0 i

di
q 1 l 1

p
d i p
0

il0 k l c lj
il lj ij
d i0 l 1
l 1

sup ij Td i x j k

1 p i ij

Td

4122
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________
n

d
p

M ij inf T il0 c lq iq a qj 1 0 i
di
q 1 l 1

p
d i p
0

il0 k l c lj
il
lj
ij

d i0 l 1
l 1

sup ij Td i x j k

d i p 0

ij k j r j k
T 0

d
j 1

Li sup
p
n

T 0 c f k
il lj
ij j

j 1 l 1

sup i Td i

d i p 0

ij k j r j k
T 0

d i j 1

M i inf
p
n

0
T
il c lj ij f j k

j 1 l 1

1 p i ij

Td

Td i

Td i

Time (sec)

Figure 6. The time response of the manifold pressure on condition A

( P0 15cm Hg ), solid line with TL 10 sin N kgw and time delay

sup i Td i
In order to simplify the switching gain choice, one can let
ij0 ij0 ij0 , ij ij ij , i i i , then the
bounds of the switching gains are
min M ij0 , ij0 ij0 max L0ij , ij0

min M
min M

ij

manifold pressure and the related idle speed under condition A

with different external load and the time-delay. Obviously, the
idle speed can be quickly adjusted back to the setting idle speed
while the different load loaded within 1.5 sec.

d 0.03

sec, dashed line with TL

10 sinN kgw and no time delay,

10 kgw and time delay d 0.03 sec, and dashed

dotted line with TL 10 kgw and no time delay

dotted line with TL

, max L ,
, max L ,
ij

ij

ij

ij

IV.

THE SIMULATION RESULTS

Usually the engine can be assumed to work under many

different operating points. The particular operating points are
chosen to valid the normal operating conditions [7]. According
to reference [7], the Fiat Dedra engine can be assumed to work
in three standard operating points. The parameters of the linear
model of the Fiat Dedra engine in three conditions are shown in
table 1.
TABLE I.

TABLE 1 THE PARAMETERS OF THE LINEAR MODEL OF THE

ENGINE

K
KP
KN 1
GP
KN 2
GA
Bm
J

Time (sec)

Figure 7. The time response of the idle speed on condition A

( N 0 600 rpm ), solid line with TL 10 sinN kgw and time

delay d 0.03 sec, dashed line with TL 10 sinN kgw and

Cond. A

Cond. B

Cond. C

2.1608

3.4329

2.1608

0.1027

0.1627

0.1027

d 0.03 sec, and dashed dotted line with TL 10 kgw and no

0.0357

0.1139

0.0357

time delay

0.5607

0.2539

0.5607

2.0183

1.7993

1.2993

4.4962

2.2047

2.22962

2.0283
1.0000

1.8201
1.0000

1.8201
1.0000

no time delay, dotted line with TL 10 kgw and time delay

The time response of the manifold pressure and the idle

speed on condition B are shown in fig. 8~9. Even the manifold
pressure seems to be changed in order to offer more power, the
idle speed still maintain the constant speed in the short time.

The controller object is to maintain the idle speed while

working under different condition and different external load
added at 5 sec. The designed controller will against the
parameter variation and the external disturbance and keeps the
robust property. The fig. 6 and fig. 7 show the response of the
4123
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________

Time (sec)

Figure 8. The time response of the manifold pressure on

condition B ( P0 15cm Hg ), solid line with TL 10 sinN

kgw and time delay d 0.03 sec, dashed line

Time (sec)

Figure 10. The time response of the manifold pressure on

condition C ( P0 15cm Hg ), solid line with TL 10 sinN

kgw and time delay d 0.03 sec, dashed line

with TL 10 sinN kgw and no time delay, dotted line

with TL 10 sinN kgw and no time delay, dotted line

with TL 10 kgw and time delay d 0.03 sec, and dashed

dotted line with TL 10 kgw and no time delay

with TL 10 kgw and time delay d 0.03 sec, and dashed

dotted line with TL 10 kgw and no time delay

Time (sec)

Figure 9. The time response of the idle speed on condition B

( N 0 600 rpm ), solid line with TL 10 sinN kgw and time

Time (sec)

Figure 11. The time response of the idle speed on condition C

( N 0 600 rpm ), solid line with TL 10 sinN kgw and time

delay d 0.03 sec, dashed line with TL 10 sinN kgw and

delay d 0.03 sec, dashed line with TL 10 sinN kgw and

no time delay, dotted line with TL 10 kgw and time delay

d 0.03 sec, and dashed dotted line with TL 10 kgw and no
time delay

no time delay, dotted line with TL 10 kgw and time delay

d 0.03 sec, and dashed dotted line with TL 10 kgw and no
time delay

The time response of condition C is shown in the following

two figures. Obviously, the idle speed is pulled back to setting
point quickly with different load added. It shows the robustness
to against the external load.

V.

CONCLUSION

The novel discrete dynamic sliding mode controller is

proposed in this paper. The detailed design procedure is also
shown, too. According to the novel existence conditions of the
discrete sliding mode, the integral gain matrix and the
switching gain matrix are chosen properly. The controller is not
only keep the motion toward the sliding surface, but also make
sure the distance from the sliding surface convergent in each
sampling time. The proposed controller is applied to the idle
speed control of the spark ignition engine. The engine usually
works under different conditions according to the
environments. The different kinds of the load are also included
due to the ramp of the road. The simulation results show the
performance of the proposed control scheme. Obviously, the
4124

IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________

International Journal on Recent and Innovation Trends in Computing and Communication

Volume: 2 Issue: 12

ISSN: 2321-8169
4118 - 4125

_______________________________________________________________________________________________
proposed controller can against the parameter variation and
external disturbance.

[8]
[9]

REFERENCES
[1]

[2]

[3]

[4]

[5]

[6]

[7]

Hazell, P.A. and Flower, J.O., Sampled-data theory applied to the

modeling and control analysis of compression ignition engines, - part I,
International Journal of Control, p.549-562, 1971.
Hazell, P.A. and Flower, J.O., Discrete modeling of spark-ignition
engines for control purposes, International Journal of Control, p. 625639, 1971.
Powell, B.K., Cook, J.A. and Grizzle, J.W., Modeling and analysis of
an inherently multi-rate sampling fuel injection engine idle speed control
loop, American Control Conference, p. 1543-1548, 1987.
Powell, B.K., Cook, J.A. and Grizzle, J.W., Modeling and analysis of an
inherently multi-rate sampling fuel injection engine idle speed control
loop, Journal of Dynamic Systems, Measurement and Control, p. 405411, 1987.
Cook, J.A. and Powell, B.K., Modeling of an internal combustion
engine for control analysis, IEEE Control Systems Magazine, p. 20-26,
1988.
Moskwa, J.J. and Hedrick, J.K., Modeling and validation of automotive
engines for control algorithm development, Journal of Dynamic
Systems, Measurement and Control, p. 278-285, 1992.
Abate, M.B., Barmish, R., Murillo-Sanchez C. and Tempo, R.,
Application of some new tools to robust stability analysis of spark
ignition engines: a case study, IEEE Transactions on Control Systems
and Technology, p. 22-30, 1994.

[10]

[11]

[12]
[13]

[14]

[15]

[16]

Nishimura, Y. and Ishii, K., Engine idle stability analysis and control,
SAE 860415, 1987.
Moskwa, J.J., Sliding mode control of automotive engines, American
Control Conference, p. 1040-1045, 1989.
Bidan, P., Boverie, S. and Chaumerliac, V., Nonlinear control of a
spark-ignition engine, IEEE Transactions on Control Systems and
Technology, p. 4-13, 1995.
Shiao, Y. and Moskwa, J.J., Cylinder pressure and combustion heat
release estimation for SI engine diagnostics using nonlinear sliding
observers, IEEE Transactions on Control Systems and Technology, p.
70-78, 1995.
Utkin, V.I., Sliding Modes in Control and Optimization,. SpringerVerlag, 1992.
Chuang, C.W., Chiang, C.L., Kao, C.C. and Fung, R.R., Applying
Modified MIMO DIVSC to Synchronism and Tension Control of Dual
Motor Systems, Journal of Dynamic Systems Measurement and
Control-Transactions of The ASME, vol. 131, no. 5, pp. 0545011~054051-7, 2009.
Chuang, C.W., Huang. C.L., Kao, C.C. and Fung, R.F.,
Synchronization and tension control of dual motor systems via MIMO
discrete pseudo model following integral variable structure control,
Mechanism and Machine Theory, vol. 44, no. 2, pp. 499-510, 2009.
Chern, T.L., Chuang, C.W. and Jiang, R.L., "Design of Discrete Integral
Variable Structure Control Systems and Its Application to Brushless DC
Motor Control," Automatica, pp 773-779, 1996.
Nise, N.S., Control systems engineering, Second Edition, New York,
The Benjamin Cummings Publishing Company, 1995.

4125
IJRITCC | December 2014, Available @ http://www.ijritcc.org

_______________________________________________________________________________________