Beruflich Dokumente
Kultur Dokumente
ﺍﻟﺨﻼﺼﺔ
ﺒﺎﻻﻀﺎﻓﺔ, (SMC) ﻫﺫﺍ ﺍﻟﺒﺤﺙ ﻴﺭﻜﺯﻋﻠﻰ ﺘﻤﺜﻴل ﻭﺘﻁﺒﻴﻕ ﺴﻴﻁﺭﺓ ﺍﻟﻨﻤﻁ ﺍﻟﻤﻨﺯﻟﻘﺔ ﺍﻟﺘﻘﻠﻴﺩﻴﺔ
Sliding mode ) ﺍﻟﻰ ﺒﻌﺽ ﺃﻻﻨﻭﺍﻉ ﻤﻥ ﺍﻟﻁﺭﻕ ﺍﻟﻤﺤﺴﻨﺔ ﺍﻟﺸﺎﺌﻌﺔ ﻟﻤﺴﻴﻁﺭﺍﺕ ﺍﻟﻨﻤﻁ ﺍﻟﻤﻨﺯﻟﻘﺔ
( ﺒﺄﺴﺘﺨﺩﺍﻡ ﺘﻘﻨﻴﺔ ﺒﺭﻤﺠﺔ ﻭﺍﻋﺎﺩﺓ ﺒﺭﻤﺠﺔ ﺍﻟﻤﺎﺩﻴﺎﺕ ﻤﺴﺘﻨﺩﺓ ﻋﻠﻰ ﺭﻗﺎﻗﺔ ﺍﻟﻤﺼﻔﻭﻓﺎﺕcontrollers
ﻭﺫﻟﻙ ﺒﺴﺒﺏ ﺍﻥ ﻫﺫﻩ ﺍﻟﺭﻗﺎﻗﺎﺕ ﻫﻲ ﺨﻴﺎﺭﺍﺕ ﺠﺫﺍﺒﺔ ﺠﺩﺍ" ﻟﺘﻨﻔﻴﺫ ﺍﻟﻜﻴﺎﻥ, (FPGA) ﺍﻟﻤﺒﺭﻤﺤﺔ
(Sliding mode) ﺍﻟﻬﺩﻑ ﺍﻻﺴﺎﺴﻲ ﻤﻥ ﻫﺫﺍ ﺍﻟﺒﺤﺙ ﻫﻭ ﺘﻨﻔﻴﺫ ﺃﺭﺒﻌﺔ ﺃﻨﻭﺍﻉ ﻤﻥ ﺍﻟﻤﺴﻴﻁﺭﺍﺕ ﻨﻭﻉ.ﺍﻟﻤﺎﺩﻱ
2494
ﻓﻲ ﻫﺫﺍ ﺍﻟﻌﻤل ﻗﺩ ﺘﻡ.( ﻭﺫﻟﻙ ﻟﻐﺭﺽ ﺍﻟﻤﺴﺎﺤﺔ ﺍﻟﻘﻠﻴﻠﺔ ﻭﺍﻟﺴﺭﻋﺔ ﺍﻟﻌﺎﻟﻴﺔ ﻟﻠﺘﻨﻔﻴﺫFPGA) ﺒﺄﺴﺘﺨﺩﺍﻡ ﺍل
( ﺒﺤﻴﺙ ﻜﺎﻨﺕ ﻨﺴﺒﺔ ﺍﻟﻤﺴﺘﺨﺩﻡ ﻤﻥXcv1000-fg680-4) ﺍﺴﺘﺨﺩﺍﻡ ﺭﻗﺎﻗﺔ ﺍﻟﻤﺼﻔﻭﻓﺎﺕ ﺍﻟﺒﺭﻤﺠﻴﺔ ﻨﻭﻉ
ﻭﻟﻐﺭﺽ ﺍﻟﺘﺄﻜﺩ ﻤﻥ ﺍﻻﺩﺍﺀ ﺍﻟﻭﻅﻴﻔﻲ.0.032 ns ﻭﺘﺄﺨﻴﺭ ﺼﺎﻓﻲ ﺃﻗﺼﺎﻩ%86 ﻫﺫﻩ ﺍﻟﺭﻗﺎﻗﺔ ﻫﻭ
( ﺒﺎﺴﺘﺨﺩﺍﻡTiming Simulation) ﺒﺼﻭﺭﺓ ﺼﺤﻴﺤﺔ ﻟﻠﻤﺴﻴﻁﺭﺍﺕ ﻓﻘﺩ ﺘﻡ ﺃﺠﺭﺍﺀ ﺍﻟﻤﺤﺎﻜﺎﺓ ﺍﻟﻭﻗﺘﻴﺔ
synthesis the data ) ( ﻭﻻﻜﻤﺎل ﺒﺎﻗﻲ ﺍﻟﻌﻤﻠﻴﺎﺕ ﺍﻟﺘﻨﻔﻴﺫﻴﺔActive-HDL simulator) ﺒﺭﻨﺎﻤﺞ ﺍل
ﻭﺃﺨﻴﺭﺍ ﻭﻟﻐﺭﺽ ﺘﺠﻴﻙ.( ISE 4.1i) ( ﻓﻘﺩ ﺘﻤﺕ ﺒﺄﺴﺘﺨﺩﺍﻡ ﺒﺭﻨﺎﻤﺞpaths and place and route
ﺍﻨﺠﺯﺕ ﻋﺩﺓ ﻤﺤﺎﻜﺎﺓ ﺤﺎﺴﻭﺒﻴﺔ ﻋﻠﻰ ﻨﻤﺎﺫﺝ ﻷﻨﻅﻤﺔ ﺨﻁﻴﺔ ﻭﻏﻴﺭ ﺨﻁﻴﺔ ﺒﺄﺴﺘﺨﺩﺍﻡ,ﻋﻤل ﻫﺫﻩ ﺍﻟﻤﺴﻴﻁﺭﺍﺕ
Mat. Lab. ﺒﺭﻨﺎﻤﺞ ﺍل
Introduction
as well as excite unmodelled and high
Sliding Mode Control (SMC) is frequency plant dynamics [8].
a technique derived from Variable There exist several techniques to
Structure Control (VSC) which were eliminate chattering. The widely-
discussed first in the Soviet literature adopted approach to chattering-free
[1, 2], and have been widely VSC is the so-called boundary layer,
developed in recent years. where the discontinuous VSC control
Comprehensive surveys of variable law signum function is applied to a
structure control can be found in [3, small vicinity around the sliding
4]. SMC is often favored basic control surface (see [4, 9]). Unfortunately,
approach, especially because the boundary layer controllers do not
insensitivity property toward the guarantee asymptotic stability but
parametric uncertainties and the rather uniform ultimate boundedness
external disturbances [5]. [10]. As a consequence, there exists a
trade-off between the smoothness of
SMC are characterized by control signals and the control
control laws that are discontinuous accuracy. Some boundary layer width
on a certain manifold in the state modification techniques to improve
space, the so-called sliding surface [2, tracking precision are discussed in [9,
6]. The control law is designed such 11]. Nevertheless, these proposed
that the representative point’s methods in practice lead to a
trajectories of the closed-loop system computational burden when
are attracted to the sliding surface and implemented, or are applicable only to
once on the sliding surface they slide linear systems [7].
towards the origin [7]. However the Another solution to chattering
major drawback in the SMC approach problem is to attempt to smooth the
is the undesired phenomenon of signum function to obtain a
chattering because of the continuous approximation. The
discontinuous change of control laws signum function is seen to be relay-
across the sliding surface. In practical like in nature. The ideal relay
engineering systems, chattering may characteristic is often impossible to
cause damage to system components, implement. One possible answer is to
2495
2496
or just entity. It has two main parts: where λ, λi are a strictly positive
the entity declaration and the constant, and e(t ) = rd (t ) − y(t ) is the
architecture [21]. In VHDL an entity
is the equivalent of an IC package in tracking errors, the rd (t ) is the
electronics. A package is defined by desired input (setpoint) and y(t) is the
its name and the number and nature of output measurement (state). while
ports it used to exchange data and e&(t ) is the time derivative of the
interact with external circuits. It is not t
2497
2498
2499
2500
((S= (S2S1) =00) and SF=00), this conventional SMC are given in the
lead to 4-bit enable vector (EN) equal Appendix A (see this Appendix A).
to 0001 that mean EN1 is active and For simplicity, we refer to the first
FI=001, therefore the first type of ESMC approach(SMC with boundary
controllers is selected and layer) as E1-SMC, The second
implemented the first sliding function ESMC (PI-sliding mode controller) as
equation for different values of error E2-SMC, while the third one (
(e) and change of error (ce), so the boundary SMC with new
control action (u) is the output of the approximation sign function) is refer
first controller(general). as E3-SMC, note that this controller
In Fig. 6 for =גּc=1, k=10, use δ s = 0.05 for all the tested
Φ=f=0.5, S2S1=01 and SF=00, this systems.
lead to 4-bit enable vector (EN) equal A. Linear Models
to 0010 that mean EN2 is active and Two second-order linear systems
FI=001, therefore the second type of examples have been tested; the first
controllers (SMC with boundary example is a system with the transfer
layer) is selected and implemented the function
first sliding function equation for
different values of error (e) and 2
G(s) = …(14)
change of error (ce), so the control s + 4s + 3
2
2501
all the ESMCs track the input without Fig. 16c shows that the control
any oscillation but with steady state signals of E1-SMC and E2-SMC are
error less than 0.6%, 0.8%, and 2.26% smooth more than the control signal
for E2-SMC, E3-SMC, and E1-SMC of E3-SMC.From the above linear
in respectively, that mean the E2- examples, it is easy to conclude that
SMC and E30-SMC are given better all the performance of ESMCs is
performance than the E1-SMC remarkable especially E2-SMC, this is
although it use smaller boundary because the structure of it contain the
thickness ( Φ = 0.2 ) than the other. integral term KI that help to improve
Furthermore, it is obvious from these the performance of it, therefore it has
results that all ESMCs regulate the steady state error less than E1-SMC
system with remove the chattering and E3-SMC.
from the control comparing with
conventional SMC which suffer from B. Nonlinear Models
this problem. Two examples are chosen here for
The second linear example has the simulation and comparison, as
following transfer function. mention in above these examples are
1 taken from [28]. The first nonlinear
G(s) = … (16) example has the following simple
s( s + 100)
mathematical model:
Which is only marginally stable,
the sliding functions and the control y& (t ) = 0.0001 y(t ) + u (t )
… (17)
parameters that used for this example
are given in Table(5). If reference is For ramp input and controller
parabolic signal r d(t)=0.15t2, then parameters given in Table (6), this
results are shown in Fig.16. This table table show that all the ESMCs use the
shows that all the SMCs use the sliding function of Eq.(2), The
sliding function of Eq.(2). We simulation results are given by Fig.
observed from Fig.16a, that all the 17, we observed from these results
SMCs approaches are able to track the that all the SMCs approaches produce
parabolic signal, it is also clearly faster convergence rate to the desired
reveals from Fig.16b that the E2-SMC ramp input without chattering and
has less than 0.0072% steady state with smaller error comparing with the
error, and the steady state error with conventional SMC, see the appendix
E1-SMC and E3-SMC is less than to show the difference between these
0.738% and 0.753% in respectively, controller. Finally the enhancement
which mean that the performance of SMC approaches have been compared
E1-SMC and E3-SMC is same but by using a another nonlinear example
model from [28], this model is
since E1-SMC use λi = 5.1 this make
described by;
the steady state error became smaller
than the E3-SMC. y&(t ) = − y(t ) + 0.5 y2 (t ) + d (t ) + u(t )
….(18)
2502
2503
[10] M. Corless and G. Leitmann: [17] Xilinx, Inc.: "Virtex 2.5 V Field
“Continuous State Feedback Programmable Gate Arrays";
Guaranteeing Uniform Ultimate available at http://www.xilinx.com.
Boundedness for Uncertain [18] Z. Navabi: “VHDL: Analysis and
Dynamic Systems”; IEEE Modeling of Digital Systems”;
Transactions on Automatic McGrow-Hill Companies, Inc.,
Control, Vol. 26, pp. 89-94, Oct. 1993.
1981.
[19] J. BHASKER: “A VHDL
[11] M. Chen, Y. Hwang, and M. Primer”; Prentice-Hall, Inc.,
Tomizuka, “A State-Dependent Third Edition, 1999.
Boundary Layer Design for
[20] M. Zwolinski: “Digital System
Sliding Mode Control ”, IEEE
Design with VHDL”; Prentice-
Transactions on Automatic
Hall, Inc., 2000.
Control, Vol. 47, No. 10, pp.
1677-1681, Oct. 2002. [21] S. Brown and Z. Vranesic:
“Fundamentals of Digital Logic
[12] Edwards C.and Spurgeon S.K.;
with VHDL Design”; McGrow-
” Sliding Mode Control: Theory
Hill Companies, Inc., 2000.
and Applications”;
Taylor&Francis, UK, 1998. [22] S. Sjoholm and L. Lindh:
“VHDL for Designers”;
[13] Goh, K.B.; Dunnigan, M.W.; Prentice-Hall, Inc., 1997.
Williams, B.W. “Robust
2504
2505
Table(3) Control actions of SMCs select. Table (7): Example (4) parameters.
U EN1 EN2 EN3 SMC type K Φ λi KI λ
U=U1 1 0 0 E1-SMC 2 0.5 --- --- 3.5
U=U2 0 1 0 E2-SMC 2 0.5 --- 0.122 3.5
U=U3 0 0 1 E3-SMC 2 0.5 0.3 --- ---
U=U4 0 0 0
2513
rd e u us y
SMC Process
+ _
-
e&
S=0
S=0
−Φ
Φ
2513
Figure. 6: SMC with boundary layer with saturation function with sliding
Function of Eq.(1) when s< Φ and when s>= Φ
Figure. 7: SMC with PI sliding control law with sliding function of Eq.(1)
2508
Figure. 10: Boundary SMC with new approximation sign function with sliding function of
2509
Eq.(1)
Figure. 11: Boundary SMC with new approximation sign function with
sliding function of Eq.(2)
2510
1 15
0.8
10
0.6
5
0.4
0 E1-SMC
E1-SMC
0.2 E2-SMC
E2-SMC
E3-SMC
E3-SMC
0 -5
0 2 4 6 8 10 0 2 4 6 8 10
Time(Se c) Time(Sec)
Fig. 15a: The output response of ESMCs for the linear Fig.16a: The output response of ESMCs with linear second
system of example(1) with a unit step input. order system of example(2) with a parabolic input.
1
0.01
E1-SMC
E2-SMC
0.8 E3-SMC 0.008
Erro r Sig na l
E r r o r S ig n a l
0.6
0.006
0.4 0.004
E1-SMC
0.2 0.002 E2-SMC
E3-SMC
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Time(Sec) Time(Sec)
Fig. 15b: Error signal curves for linear system Fig. 16b: Error signal curves for linear system
of example(1) with ESMCs. of example(2) with ESMCs.
7
350
E1-SMC
6 E2-SMC
E3-SMC
300
Control Signal
5
250
C o n tr o l S ig n a l
4 200
3 150
2 100 E1-SMC
E2-SMC
1 50 E3-SMC
0 2 4 6 8 10
Time(Sec) 0
0 2 4 6 8 10
Time(Sec)
Fig. 16c: Control signal curves for linear
Fig. 15c: Control signal curves for linear system system of example(2) with ESMCs.
of example(1) with ESMCs.
2511
12 1
10
0.8
O utp o u t R e s po n se
O u t p o u t R e sp o n se
8
0.6
6
0.4
4
E1-SMC
E2-SMC 0.2 E1-SMC
2
E3-SMC E2-SMC
E3-SMC
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Time(Sec) Time(Sec)
Fig. 17a: The output response of ESMCs with Fig.18a: The output response of ESMCs with
nonlinear system of example(3) with a ramp input. nonlinear system of example(4) with a unit step input.
0.07 1
E1-SMC
0.06 E1-SMC
E2-SMC
0.8 E2-SMC
0.05 E3-SMC
E3-SMC
Error Signal
E r r o r S ig n a l
0.04
0.6
0.03
0.02 0.4
0.01
0.2
0
-0.01 0
0 2 4 6 8 10 0 2 4 6 8 10
Time(Sec) Time(Sec)
Fig. 17b: Error signal curves for nonlinear Fig. 18b: Error signal curves for nonlinear
system of example(3) with ESMCs. system of example(4) with ESMCs.
1.4 2
1.2 1.8 E1-SMC
E2-SMC
1 1.6 E3-SMC
C o ntr ol SIg na l
C o n t r o l S ig n a l
0.8 1.4
1.2
0.6
1
0.4 E1-SMC
E2-SMC 0.8
0.2 E3-SMC
0.6
0
0 2 4 6 8 10 0.4
Time(Sec) 0 2 4 6 8 10
Fig. 17c: Control signal curves for nonlinear Time(Sec)
system of example(3) with ESMCs. Fig. 18c: Control signal curves for nonlinear
system of example(4) with ESMCs.
2512
Appendix A
This appendix illustrate the simulation results (output response, error signal, and
control signal) of the four examples that are given in section VI when the control
system uses the conventional SMC(u=Ksgn(S)) with sliding function of Eq.(1) for
examples 1,2,4 and with sliding function of Eq.(2) for example 3. The inputs and
the control parameters with this controller for each example are given in Table
(A.1), the values of K parameters are same the values of K in the ESMCs
approaches except in example 2 and 4. Also this appendix illustrate the response
of the tested example (with unity feedback) without controller.
Table (A.1): input and parameters for the simulation examples .
Example Type of input K λ λi
No.
Ex.1 Step input 6.5 2 ---
Ex.2 Parabolic input -400 -6 ---
Ex.3 Ramp input 8 --- 0.1
Ex.4 Step input -2 -8 ---
The results for the simulation examples are shown in the following figures, where
the dished line ’---‘ refer to response of these examples without controller, while
the sold line ’__’ refer to response of these examples with the conventional SMC.
2513
Appendix B
right half plane, the second one is linear time varying model
y& (t ) = a (t ) y(t ) + b(t )u (t ) , where a(t)=0.001t and b(t)=0.15t, t is varying
from zero to 10Sec. The sliding functions, types of controllers and the
control parameters that are used for controlling these examples are given in
Table(B.1) and Table(B.2). This appendix also illustrates the response of
these simulated examples without controller (with unity feedback).
Table (B.1): types of controllers and control Table (B.2): types of controllers and control
parameters for the linear unstable example. parameters for the linear time varying example.
SMC type K Φ λi KI λ SMC type K Φ λi KI λ
2314
If reference is unit step signal, then the results for the simulation examples
are shown in the following figures. As in appendix A, in the figures
without legend, the dished line ’---‘ refer to response of these examples
without controller, while the sold line ’__’ refer to response of these
2315
2316