Documente Academic
Documente Profesional
Documente Cultură
e and det( ) 0 C B
= . The
sliding mode control
( )
1
( ) sgn( )
l n
u u u C B C Ax K
= + = + (2)
where
( )
1 2
diag , , ,
m
K K K K
= =
The system in the sliding mode (the reduced-order system) is
obtained by substituting
eq
u
in (1)
( )
1
( ) x I B C B C Ax
= (3)
The m eigenvalues of
1
( ) I B C B C
C
is
selected such that the n-m remaining eigenvalues lie in the left
half-plane. If the reduced order system (3) is stable then the
stability of the original system (1) is guaranteed. Therefore,
this selection is required for stability of the original system
(1).
The sliding mode control is insensitive to matched
uncertainties. In fact, the matched disturbances do not affect
the system in the sliding mode (3). A drawback of SMC is the
chattering resulting from discontinuous control. There are
many methods to reduce the chattering including continuous
approximations. One can consider the following continuous
approximation for the nonlinear term of the control
, 0 1, 1,
i
ni i
i i
u K i m
= < < =
+
(4)
where
ni
u indicates the i-th element of the vector control
input [2]. Another suitable continuous approximation is
1 if
( ) if | |
-1 if
i i
ni i i i i i
i i
u K
>
= s
<
(5)
where
i
is a small positive real number and ( )
i i
is a
continuous function crossing the origin and at points
( , 1)
i
[2, 9]. A simple and suitable choice is
( )
i
i i
i
s
=
.
The dynamic sliding mode control (DSMC) is an alternative
method for reducing the chattering [13, 14]. The integral of
controller (2) with the nonlinear term (4) or (5) as a new
control, yields the dynamic behaviour. The system may be
stabilised via DSMC with an appropriate sufficient condition
[14].
3 PID controllers
PID controllers are dominant and popular and, have been
widely used since the 1940s because one can obtain the
desired system responses and it can control a wide class of
systems. This may lead to the thought that the PID controllers
give solutions to all requirements, but unfortunately, this is
not always true [15]. Tuning PID controllers have been a
challenging problem since its appearance (Ziegler-Nichols,
1942) in control engineering. Alternative tuning methods
have been recently presented including disturbance rejection
magnitude optimum [16, 17], pole placement and
optimization methods [18, 19]. These methods provide
relatively fast and non-oscillatory disturbance rejection
responses. In particular, these methods do not require any
additional tuning parameters.
General valuation and a design method to find the optimal
parameter setting for a controller of given structure has been
studied in [19]. In fact, for various plants optimal PI and PID
controllers are designed by optimising low frequency
performance. The trade-off between performance, robustness
and control activity is regarded an important issue which one
can select tuning parameters.
A method has been presented to find an optimal controller
for a practical definition of optimality which enables
highlighting the effects of the gain margin, complementary
sensitivity bound, low frequency sensitivity and high
frequency sensor noise amplification [19]. In this method, the
two parameters of PI controllers satisfying the constraints
correspond to a given domain in a plane. The optimal
controller lies on the curve. The design plot enables
identification of the PI controller for desired robust
conditions, and in particular, gives the PI controller for lowest
sensitivity. By applying this method, trade-off among high-
frequency sensor noise, low frequency sensitivity, gain and
phase margin constraints are also directly available.
The transfer function of a PID controller is given by
1
( ) (1 )
P D
I
K s K T s
T s
= + + (6)
where
P
K ,
P
I
K
T
and
P D
K T
represent the proportional,
integral and derivative gains of the controller, respectively.
Define
1
n
I D
T T
=
and
1
2
I
D
T
T
=
as the controllers natural
frequency and the damping coefficient, respectively. Then the
PID transfer function (6) can be written as
2 2
2
( )
2
n n
p
n
s s
K s K
s
+ +
=
(7)
3 PISMC design
Proportional integral sliding mode control (PISMC) is
basically obtained by adding an integral term to the sliding
function which provides an extra dynamic to the system and
more flexibility for tuning and obtaining desired performance.
PISMC is more general than the classical SMC or
proportional SMC (PSMC). If PSMC design is only required,
the integral term is excluded. In many cases, a PSMC may
yield desired system responses. Sometimes for obtaining the
desired performance, an integral SMC (ISMC) may be
required. However, the combination of PSMC and ISMC
benefits from the advantages of both approaches and may give
the desired system responses.
The sliding function is defined as
Cx P xdt
= +
(8)
where
m n
P R
e . Consider the system (1) with ( , ) 0 f t x =
and define the sliding surface
0 Cx P xdt
= + =
(9)
The gain matrix C is selected such that det( ) 0 CB = . Then
the equivalent control is
( ) ( )
1
eq
u CB CA P x
= + (10)
Now consider the Lyapunov function
1
2
T
V = (11)
Then
( )
T T
V C Ax Bu P x
= = + +
(12)
V
is negative definite if
( ) ( ) ( ) ( )
1 2
sgn( )
T
n
C Ax t Bu t P x k k k
+ + s + + +
where 0, 1, ,
i
k i n > = . So the following control guarantees
the system trajectories reaches the sliding surface (8) in finite
time
( ) ( )
1
sign( ) u CB CA P x K s
= + +
(13)
where
( )
1 2
diag , , ,
n
K k k k = . The design matrices C and
P
= + +
= +
=
=
(14)
with the following physical parameters:
a
E : The input terminal voltage (source), (v);
SMC system
Reference
signal
Sliding
function
SMC
DC
Motor
+
Output
E
b
DC
Motor
Inertia
load J
T
Torque
Load
E
a
L
a
Damping B
R
a
b
E : The back emf, (v);
a
R : The armature resistance, (ohm);
a
I : The armature current (Amp);
a
L : The armature inductance, (H);
J : The moment inertial of the motor rotor and load,
(kg.m
2
/s
2
);
T : The motor torque, (Nm)
: The speed of the shaft and the load (angular velocity),
(rad/s);
: The shaft position, (rad);
B : The damping ratio of the mechanical system, (Nms);
T
K : The torque factor constant, (Nm/Amp);
b
K : The motor constant (v-s/rad).
Assume that
T b
K K K = = . A state space representation of
(14) is
0
1
a
a
a a
a
a a
B K
J J
E
R K I I
L
L L
= +
(15)
Let
=
1
x ,
2 a
x I = ,
1
B
a
J
= ,
2
K
a
J
= ,
3
a
K
a
L
= ,
4
a
a
R
a
L
= ,
1
a
b
L
= ,
a
u E =
Then the system (15) can be written as
1 1 1 2 2
2 3 1 4 4
1
x a x a x
x a x a x bu
y x
= +
= + +
=
(16)
where y is the output of the system. To obtain a simple
sliding mode system (the so-called reduced-order system), the
system (16) is converted to a canonical form. Obviously, the
results may be represented in terms of the actual physical
states of (15). To this end assume that
1 1
1 2
z x
z z
=
=
(17)
Then the system can be converted to a the following canonical
form
1 2
2 1 1 2 2
1
z z
z a z a z bu
y z
=
= + +
=
(18)
where
1 2 3 1 4
a a a a a = ,
2 1 2
a a a = + and
2
b a b =
. Select the
sliding surface
( )
1 2
0 c r z z = + = (19)
where 0 c < . The sliding mode control is
sgn ) u K ( = (20)
where 0 K > is selected sufficiently large to enforce the
system trajectory to remain on the sliding surface.
The equivalent control (the control during the sliding
mode) is then given by
( )
1 1 2 2
eq
a z c a z
u
b
+ +
=
(21)
When 0 = , ( )
2 1
z c r z = . The system in the sliding mode
is
( )
1 1
d
r z z
dt
= (22)
From (19) and (22)
( ) ( )
1 1
d
r z c r z
dt
= (23)
which represents the reduced-order system, and because
0 < c then this system is asymptotically stable, i.e.
1
lim
t
z r
= . Since
1
z = then lim
t
r
= .
The simulation results of the system (15) with control (20)
are shown in Figures 3 and 4. Figure 3 shows that the
response of the DC motor with SMC whilst Figure 4 depicts
the speed of the DC motor using the SMC and PID
controllers. Applying a PID controller to the system when
there is no disturbance, the output completely tracks the
desired speed, but with the traditional SMC, the speed
achieved is slightly lower than the desired speed. However,
the settling time for SMC is shorter (0.2 sec) than for PID (0.7
sec). There is no overshoot when SMC control is applied
whilst using PID controller, the speed significantly increases
during 0.2 and 0.4 period of time. To remove this obstacle the
PISMC is applied to the system.
Figure 3: The responses of the DC motor system (18) using
the SMC (20) without disturbance.
Figure 4: The output regulation for the DC motor (15) without
disturbance using PID controller and SMC.
The simulation result of applying the PID controller which
its parameters has been tuned, using the optimisation method,
are shown in Figure 5. The PID parameters are
8 1 25
I
K , T . = = and 0.18
D
T = . The simulation results (see
Figure 5) show the PID has been tuned with the following
specifications: (i) there is no overshoot; (ii) the steady state
error tends to zero; (iii) the rise time is minimised.
Figure 5: The responses of the closed-loop DC motor system
using a PID controller tuned by an optimisation technique.
Figures 6-8 show the system responses using a PID
controller and the classical SMC with matched disturbance.
SMC is insensitive against the (matched) disturbance and the
desired speed is obtained even in the presence of disturbance.
A well-tuned PID controller may reduce the affect of the
disturbance on the system. However, by applying PID
controllers, it is normally impossible to reject the disturbance
completely.
Figure 6: The phase plane, the disturbance and the sliding
function for the system.
5 Control of a DC motor with
disturbance using PISMC
Let
| |
1 2
P P P