Sunteți pe pagina 1din 6

A Comparison of Robustness: Fuzzy Logic, PID, & Sliding Mode Control

Charles P. Coleman and Datta Godbole Department of Electrical Engineering and Computer Sciences University of California at Berkeley Berkeley, CA 94720, USA

Abstract This work is performed in order to foster and promote unbiased and accurate comparison of fuzzy logic control and classical control design methodologies. We are motivated to execute this study by the demonstrated robustness to plant perturbations of the fuzzy logic controller given in 0]. Robust fuzzy logic, PID, and sliding mode controllers are designed to control the speed of a nominal third order linear time-invariant model of a motor. The step response performance of each controller, applied to the nominal and two perturbed motor plants, is presented. We conclude fuzzy logic control can be a useful tool for the control engineer. We encourage more benchmark comparisons of fuzzy logic control with classical control techniques for the bene t of the practicing control engineer.
The successful use of fuzzy logic controllers has greatly expanded in the last twenty years ( 0]{ 0]). This expansion has prompted much comparison to classical control techniques. The ensuing discussions have not always been amicable. Nor have they necessarily lead to manifest results and conclusions for the practicing control engineer. At times the debate over the use of fuzzy logic control techniques versus classical control techniques has become quite heated with ardent detractors on both sides ( 0]{ 0]). Some control engineers have been placed in a dilemma by this debate, and have been left without an appeasing study or comparison upon which to formulate a useful opinion.
C. Well-Being of the Control Engineer B. Debate and Dilemma A. New Control Tools, New Enemies?
I. Introduction

D. A Call for Peace or at Least a Cease Fire! and Negotiations

Fortunately, some have called for control engineers to consider inclusion of fuzzy logic control into their control enginering toolboxes, and for the utilization of both fuzzy logic control and classical control when and where appropriate ( 0] 0]). This appears to be a reasonable request. At the very least, the fuzzy logic tool can be compared to classical control tools. Should the fuzzy logic control tool prove useful and acceptable to the control engineer, it should become welcome addition to her control toolbox. The control engineer is concerned about such properties as the robustness of a controller to plant perturbations and uncertainty. We take the perspective of the control engineer who has several control tools available for the synthesis of robust controllers, and who is interested in determining the capabilities of each control tool to synthesize a robust controller. To assess the usefulness of each tool in designing a robust controller, we choose a benchmark problem, and then engage in a comparison of the control designs which result from the application of each control design method. Fuzzy logic control, PID control, and sliding mode control tools are chosen for our investigation. Fuzzy logic control is chosen because of its empirically demonstrated robustness properties shown in 0]. PID control is chosen because it is one of the most commonly used controllers in industry, and because it has plant perturbation robustness properties which can be mathematically analyzed. We choose a sliding mode controller because it is a robust non-linear control method whose robustness properties can also be mathematically analyzed. We do not claim to investigate and present every robust control technique. Indeed, the popular and successful H1 and adaptive control techniques are noticeably absent from this study. We felt it was necessary to restrict ourselves to fuzzy logic control and two other control methods, in order to keep our
II. Choice of Control Tools

E. Robust Controllers

In the process of this raging and seemingly growing battle, it appears that the best interest of the control engineer has been occluded. The practicing control engineer usually has little interest in ideological debates. It is her interest to produce a working controller, with e cacy, making use of any and all techniques at her disposal.
Research supported in part by ARO under grant number DAAL-91-G-091 and an AASERT supplement.

study brief, intelligible, and meaningful. Motivated by the work in 0] and 0], and encouraged by the robustness results presented for fuzzy logic control, we consider the robust speed control of a rotating motor, modeled by a third order linear time-invariant transfer function. The general conguration for the motor speed control problem is given in Figure 1.
r + e c
Controller

A. Motor Speed Control

III. Control Problem

loop unit step responses of the nominal plant and the two perturbed plants. To be acceptable, the closed loop step responses must simultaneously have short rise time and no overshoot. We design the fuzzy, PID, and sliding mode controllers based on the nominal plant given in equation (1). Each designed controller is then applied to all three transfer functions, and numerical simulations are performed to analyze the controller's robustness. The design of the fuzzy logic controller is given in Section IV. The design of the PID controller is given in Section V. The design of the sliding mode controller is given in Section VI. Plots of each controller's closed loop step response and control e ort are shown in Appendix A. We implement the robust fuzzy controller given in 0] and 0]. The inputs to the fuzzy controller are e and ! _ c . The output of the fuzzy controller is u. The universes of discourse of e, ! _ c , and u are partitioned into seven fuzzy sets: NB - negative big NM - negative medium NS - negative small ZE - zero PS - positive small PM - positive medium PB - positive big Each fuzzy set is represented by a Gaussian membership function. The rule base of the fuzzy logic controller contains forty-nine rules which are tabulated in Figure 2. The output of each rule is determined by min-inference. The crisp output u of the fuzzy logic controller is generated by centroid defuzzi cation.
PB PM PS ZE NS NM NB NB NB NB NB NB NM NS ZE NM NB NB NB NM NS ZE PS NS ZE NB NB NB NM NM NS NS ZE ZE PS PS PM PM PB
e

F. Design Approach

Motor Plant

Figure 1: Motor Speed Control Problem We use the nominalplant, perturbed motor plants, and the performance criteria given in 0] to design and compare robust fuzzy, PID, and sliding mode controllers. As shown in Figure 1, the reference step input speed !r and the output motor speed !c are available for comparison. Each controller only has access to the motor speed error e = !r ?!c and the motor angular acceleration ! _ c as inputs. Each controller produces only one output u.
C. Nominal Plant B. Restrictions on Controller Inputs and Outputs

IV. Analysis and Reconstruction of a Fuzzy Robust Controller

The nominal motor plant is modeled by the following transfer function: 5 G1(s) = s(s + 1)( s + 2)
D. Perturbed Plants

(1)

The two perturbed motor plants are given by the following transfer functions: 15 G2(s) = s(s + 1)( s + 2) 5 G3(s) = s2 (s + 2)
E. Performance Criteria

(2) (3)

! _c

PS PM NM NS NS ZE ZE PS PS PM PM PB PB PB PB PB

PB ZE PS PM PB PB PB PB

Figure 2: Fuzzy Logic Controller Rule Base

The step response of this fuzzy logic controller apThe designed controllers must be robust to variaplied to the nominal and perturbed plants is shown tions of system parameters. Speci cally, for the in Appendix A. The step responses for all three fuzzy logic, PID, and sliding control designs, one plants have short rise times and no overshoot. Thus, single controller must render acceptable the closed

they meet the speci ed robust performance criteria. We consider the following sliding surface for this The control e ort generated by the fuzzy logic con- design:1 troller is also shown in Appendix A. S = (y ? yd ) + 1 (y _?y _d ) + 2 (y ? yd ) (7) V. Analysis and Synthesis of a Robust PID The choice of sliding surface is based on the followController ing considerations: The fuzzy logic controller designed in 0, 0] is similar to a nonlinear PD controller 0]. We design a linear The relative degree of the system (6) with S as PID controller to obtain satisfactory step response output should be 1. This ensures that the input _ u appears explicitly on the right hand side of S for all three transfer functions. equation. A linear PID controller can be characterized by the following transfer function: System dynamics on the sliding surface should be stable. (This requires 1 ; 2 > 0) i C (s) = Kp + Kd s + K (4) yd (t) speci es the reference trajectory to be tracked s by output. If one knows the entire trajectory of yd , Because, it is very di cult to realize a pure di er- then the derivative information y _d (t); yd (t) can be entiator, we use the following transfer function for extracted o line and used in the design to improve implementation. the performance. If the information is not available (or as in this case, yd (t) being unit step, is nondifKi d s ferentiable) we can assume y _d and yd to be zero and + (5) C (s) = Kp + K s+1 s let the robustness property of the sliding mode conTo have a fair comparison, we use a small value for troller take care of the mismatch. Ki = 0:001, just enough to keep steady state error With the de nition of sliding surface as above, we zero. We use a small value for = 0:01, so that its e ect on the dynamics is minimal, but the derivative have reduced the design requirement from tracking yd (t) to being on the surface S = 0. Once on the controller is still realizable. The selection of Kp and Kd is based on the root sliding surface, the dynamics (equation 7) is expolocus of C (s)G1(s). The root locus gives us the nentially stable and asymptotic trajectory tracking locations of closed loop poles. The proportional is achieved. The control u is designed to make the and derivative gain values are chosen such that the surface S = 0 attractive and to reach the surface in closed loop pole locations are in the left half com- nite time. plex plain for all three plants. Another criterion for Consider the following lyapunov function: selection of these gains is the step response of the closed loop system. 1 S2 (8) V=2 We select the following gains: Kp = 2; Kd = 5 With these gains, the closed loop system is stable for Its time derivative is given by the open loop gain up to 51. Thus in particular, the _ =S S _ V (9) above PID controller stabilizes the systems given by the transfer functions (1,2). For the model of (3), As the system has relative degree 1 with S as output, we have closed loop stability for the open loop gain we can solve for u from the equation upto 30. The step responses have no overshoot and _ = ?K sgn(S ) are critically damped. S (10) VI. Analysis and Synthesis of Robust _ and also guarThis results in a negative de nite V Sliding Mode Controller entees nite time convergence to the sliding surface. To design a sliding mode controller, we convert the But, the controller will result in high frequency chattransfer function model into state space format. The tering near sliding surface. To avoid this, we use the controllable canonical form realization of the model following expression to solve for u.2 of (1) is given by: _ = ?K S S (11) 2 3 2 3 0 1 0 0 _ = ?K S 2 ) V (12) x _ = 4 0 0 1 5 x+4 0 5u 1 Refer to 0] for details of sliding mode controller design. 0 ?2 ?3 5 2 With this controller, the trajectory is not guarenteed to reach y = 1 0 0 x (6) the sliding surface in nite time.

Expanding both sides of the equation, we get _ = ?2x2 ? 3x3 + 5u + 1 x3 + 2 x2 S = ?K (x3 + 1 x2 + 2 x1 ? 2 yd ) Desired The u obtained from this equation will be used for the control. This controller needs access to all three states, whereas the fuzzy logic controller of 0, 0] make use of only x1 and x2 . To have a fair comparison, we will construct an observer to get an estimate of the state variable x3.

results. Software tools were also available for the design and simulation of the PID and sliding mode controllers for this problems, and so these two controllers were easy to implement.
B. Analytical Tools for the Analysis of Fuzzy Logic Control

A control engineer adding the fuzzy logic control tool to her toolbox already full of classical control tools, might ask \Why does this method demonstrate robustness?", and \When will it fail?" These questions often have mathematically analytical answers for classical control tools. To satisfy this inquiry, we call for the continued development of mathStandard Luenburger observer is designed to get ematically analytical tools to answer such questions an estimate of x3 . We use the model of (1) to design to the satisfaction of the control engineering comthe observer as follows: munity. _ = Ax x ^ ^ + bu + L(y ? y ^) (13) C. Fuzzy Logic Control: A Useful Tool y ^ = cx ^ From our brief study, we conclude that fuzzy logic control should have a place in the control engineer's where, the matrices A, b and c correspond to the toolbox. Only time, experience, and further analmodel in equation (1). The matrix L is chosen such ysis will determine whether fuzzy logic control bethat the eigenvalues of observer error dynamics (A- comes a prominent tool in the control toolbox used LC) are in the left half plane and the error dynamics by devotees of classical control techniques. is faster than the dynamics on the sliding surface. VIII. Future Work The closed loop step responses with this controller are shown in Appendix A. The controller is robust We encourage more extensive fair and unbiased benchto input gain changes as well as to the changes in mark comparisons of fuzzy logic control with classisystem dynamics while keeping the input magnitude cal control techniques for the bene t of the practicsmall. ing control engineer, and we hope to become a part Note: With the use of equation (11), the con- of this e ort. troller is essentially a linear state feedback controller. In the future, we hope to contribute to the creIt results in closed loop eigenvalues of -1.3392 & ation of new analytical tools for the analysis of fuzzy -4.33 j 1.91 for model (1). The same controller, logic controllers. We also hope to be able to conwhen applied to other plants will give the following tribute to the continuing e ort to identify classes of closed loop eigenvalue locations: problems where rule based fuzzy logic control techModel of (2) ) ?1:32; ?3:55; ?19:12 niques have advantages over and are more appropriThus, by design, the increase in gain makes the ate than classical control techniques. closed loop system more stable... IX. Acknowledgments Model of (3) ) ?1:4; ?3:79 j 2:62 The authors would like to thank Professor S. Shankar VII. Conclusions Sastry for his gracious support and encouragement. A. Comments on Controller Designs The authors would also like to thank Dr. Shahram The control engineer pro cient in PID and sliding M. Shahruz for helpful discussions. mode control techniques can readily synthesize roReferences bust controllers to perform the task dictated by this example. As demonstrated in 0] and veri ed in this 1] H. T. Nguyen, C.-W. Tao, W. E. Thompson, \An Empirical Study of Robustness of Fuzzy study, the control engineer utilizing fuzzy control Systems", Proc. of 2nd IEEE Intl. Conf. on techniques can readily achieve the same goal. Thus, Fuzzy Systems , pp. 1340{1345. all three control tools are available to the control en2] R. M Tong, \An Annotated Bibliography gineer for use in synthesizing an acceptable robust of Fuzzy Control", in Industrial Applications controller for this control problem. of Fuzzy Control , M. Sugeno, Ed., NorthWith the appropriate software tools available, the Holland, Amsterdam, Holland, 1985, pp249{ fuzzy logic controller for this problem was relatively 269. simple to implement, and it provided satisfactory

3] 4] 5]

6]

7] 8] 9] 10] 11] 12] 13]

Closed loop Step Response of System 2

L. A. Zadeh, Eds., Kluwer Academic Publishers, Boston, MA, USA, 1992, pp 69{96. S. Chiu, S. Chand, D. Moore, A. Chaudhary, \Fuzzy Logic for Control of Roll and Moment for a Flexible Wing Aircraft", IEEE Control Systems Magazine, pp 42{48, Vol 11, No 4, June 1991. A. L. Schwartz, \Comments on Fuzzy Logic Control of Roll and Moment for a Flexible Wing Aircraft", IEEE Control Systems Magazine, pp 61-62, Vol 12, No 1, February 1992. S. Chiu, \Author's Reply", IEEE Control Systems Magazine, pp 62{63, Vol 12, No 1, February 1992. E. Cox, \Adaptive Fuzzy Systems", IEEE Spectrum, pp 27{31, Vol 30, No 2, February Figure 3: Closed Loop Step Response of nominal 1993. plant C. J. Herget, Ed., \Reader's Forum", IEEE Control Systems Magazine, pp 5{7, Vol 13, No 3, June 1993. M. Athans, \Control - The Adventure Continues", Bode Lecture, 32nd IEEE Conference on Decision and Control, San Antonio, TX, USA, December 15{17, 1993. E. H. Mamdani, \Twenty Years of Fuzzy Control: Experiences Gained and Lessons Learnt", Proc. of 2nd IEEE Intl. Conf. on Fuzzy Systems, pp. 339{344. M. Tomizuka, \Fuzzy Control in Control Engineer's Tool Box", Lecture, ARO/NASA
Closed loop Step Response of System 1

An Introduction to Fuzzy Logic Applications in Intelligent Systems, R. R. Yager and

Boston, MA, USA. W. Pedrycz. Fuzzy Control and Fuzzy Systems, 2nd extended ed. John Wiley & Sons, New York, NY, USA. H. Berenji, \Fuzzy Logic Controllers", in

T. Terano, K. Asai, M. Sugeno. Fuzzy Systems Theory and Its Applications. Academic Press,

A Controller Performance Results


1.2
Fuzzy Logic Controller PID Controller Sliding Mode Controller

0.8

0.6

0.4

0.2

10

15

20

25

30

Time

1.2

Fuzzy Logic Controller PID Controller

Sliding Mode Controller

0.8

0.6

September 30 { October 2, 1993. 14] C.-W. Tao, R. Mamlook, W. E. Thompson, \Reduction of Complexity for a Robust Fuzzy Controller", Proc. of 2nd IEEE Intl. Conf. on Fuzzy Systems, pp. 1346{1349. 15] H. Ying, W. Siler, J. J. Buckley, \Fuzzy Control Theory: A Nonlinear Case", Automatica, Vol. 26, No. 3, pp 513{420, 1990. 16] J.-J. E. Slotine and W. Li, Applied Nonlinear Control. Prentice Hall, 2nd ed., 1991. Figure 4: Closed Loop Step Response of the perturbed model G2(s)
0.4 0.2 0 0 5 10 15 20 25 30 Time

Workshop on Formal Models for Intelligent Control, MIT, Cambridge, MA 02139, USA,

1.2

Fuzzy Logic Controller

2
PID Controller Sliding Mode Controller

System1 (nominal) System2 System3

1.5

Closed loop Step Response of System 3

0.8

Control Effort for PID Controller

0.5

0.6

0.4 -0.5

0.2 -1

0 0 5 10 15 Time 20 25 30

-1.5 0 5 10 15 Time 20 25 30

Figure 5: Closed Loop Step Response of the perturbed model G3(s)


2
System1 (nominal) System2 System3

Figure 7: PID Controller Control E ort


3
System1 (nominal) System2 System3

1.5

Control Effort for Fuzzy Logic Controller

0.5

Control Effort for Sliding Mode Controller

-1

-0.5

-2 -1

-1.5 0 5 10 15 Time 20 25 30

-3 0 5 10 15 Time 20 25 30

Figure 6: Fuzzy Logic Controller Control E ort

Figure 8: Sliding Mode Controller Control E ort

S-ar putea să vă placă și