Sie sind auf Seite 1von 5

2010 International Conference on Computer Applications and Industrial Electronics (ICCAIE 2010), December 5-7, 2010, Kuala Lumpur,

Malaysia

Fuzzy Self-Tuning Gain Scheduled Control Design for an Autopilot Missile


Sajad Tabatabaei1,*, Shahab-o-din Tohidi2, Mokhtar Sha Sadeghi2, Parinaz Sadat Mirjafari2
1,*

Control Engineering Department, Electrical Engineering Faculty, Mahshahr Islamic Azad University, Email: sdtabatabaei@gmail.com 2 Control Engineering Department, Electrical Engineering Faculty, Shiraz University of Technology

AbstractThe paper presents the lateral acceleration control design of a missile model in which the parameters of fuzzy logic are designed to optimize the performance characteristics of the plant. The control scheme utilizes three fuzzy controllers to achieve the performances for lateral acceleration. The control scheme is based on classical compensator which its parameters are self-tuned by fuzzy logic. The simulation results show that the designed fuzzy selftuning gain scheduled controller has excellent performance. KeywordsMissile control; Gain Scheduled Control; Fuzzy Control, Lateral Acceleration Control

I. INTRODUCTION The tracking performance of a missile is dependent on the location of the flight envelope and varies with factors such as Mach number and incidence. During recent decades, some control schemes have been proposed. Several approaches, including adaptive control [1], nonlinear control [2-3], H optimal control [4] and gain scheduling [5-6] have been used to alleviate tracking problems. One of the most popular methods for applying linear time invariant control theory to time-varying and/or quasi-linear [7] systems is gain scheduling [5]. During the 1980s, Rogh and his colleagues developed an analytical framework for gain scheduling using extended linearization [6]. Gain scheduling is a well established control technology for controlling the non-linear systems [8-9]. Its application to uncertain systems is still underdevelopment. This strategy involves obtaining Taylor linearized models for the plant to satisfy local performance objective for each point and then adjusting (scheduling) the controller gain in real time as the operating conditions vary. This approach has been applied successfully for many years, particularly for aircraft and process control problems. Evolutionary algorithms [5] are also suitable alternative technique for optimal controller parameter tuning. The use of fuzzy logic control is motivated by the need to deal with highly nonlinear flight control and performance robustness problems. It is well known that fuzzy logic is much closer to human decision making than traditional logical systems. Fuzzy control based on fuzzy logic provides a new design paradigm such that a controller can be designed for complex, ill-defined

Figure 1. Magnetization as a function of applied field. Note how the caption is centered in the column [10].

processes without knowledge of quantitative data regarding the input-output relations, which are otherwise required by conventional approaches. Fuzzy logic scheduling of controller parameters is found simple and effective in a large number of applications. In this study, the fuzzy gain schedule is used to control the lateral acceleration of an autopilot missile for the best tracking in all ranges of inputs from zero to infinite. Because in missile control, there are large ranges of inputs and the missile should have capability to track all of these inputs. The rest of the paper is organized as following: Problem formulation and equations of motion are explained in Section II. The controller design approach is described in Section III. Simulation results and discussion is brought in Section IV and finally a brief discussion is done in Section V. II. PROBLEM STATEMENT The missile model used in this study is taken from Hortons M.Sc. thesis [5]. It describes a 5 degree-offreedom (DOF) model in parametric format with nonlinear behavior. In this study, the airframe is roll stabilized. The problem of controlling a missile in roll are relatively minor and the roll stabilized airframe ( = 45 ) imparts little cross-coupling between axes. Therefore, this study looks at the reduced problem of a controller for the uncoupled lateral motion (on the xy-plane in Fig. 1).

978-1-4244-9053-0/10/$26.00 2010 IEEE

196

Missile autopilots are usually designed using linear models of nonlinear equations of aerodynamic forces and moments. The objective of this paper is robust design of lateral acceleration autopilot for a nonlinear missile model. The aerodynamic parameters in this model are derived from wind-tunnel measurements. The equations of motion are given by:
v = yvv + y Ur =
i

the pitch (yaw) rate are measured respectively by accelerometer and rate gyro, and are feedback signals to the controller. The plant to be controlled is time-variant which makes the controller design difficult.

B. Accelerometer: The accelerometer is placed at a position displaced


Table I. [5] Physical Parameters Symbol a Meaning Speed of sound Air density Reference diameter Reference area Mass Lateral inertia Centre of gravity Centre of pressure Fin centre of pressure Static margin Fin moment arm Table II. [5] Aerodynamic Derivative Corresponding Force or Moment Side force Aerodynamic Derivative Interpolated Formula
2 6 + 1 .5 M 6 0
1 0 1 .4 M + 1 .5

1 VS (Cyv v + VCy ) Ur 2m
1 1 V Sd (C nv + dC nr r +V C n ) 2I z 2

(1) (2) (3)

r = nv v + n r r + n =

Value 340m/s 1.23kg/m3 0.2 m 0.0314 m2 150 kgfull 100 kg all burnt 75 kg.m2 full 60 kg. m2all burnt 1.3+ m/50 1.3 + 0.1M +0.3 2.6 m
x
x
c g

a y = v+ Ur

d S m
lz
x
cg

where V is the lateral velocity, r is the body rate, is rudder fin deflection, U is the forward velocity, a y is the lateral acceleration at the centre of gravity y v , y are semi-nondimensional force derivatives due to lateral velocity and fin deflection. nv , n r , and n are seminondimensional force derivatives due to lateral velocity, body rate and fin deflection. The aerodynamic derivative C y v , C y , C nv , C n r , and C n are the function of Mach number ( M ) and incidence angle (~ v /U for U >> v ). They are evaluated by interpolating the discrete data points obtained from the wind tunnel experiments. A. Autopilot configuration: Applying the Laplace transform to the state and output equations (1)-(3) gives body rate and acceleration transfer functions of n s (n y v nv y ) (4) Pr (s ) = 2 s ( y v + n r )s + (Unv + y v n r )
Pay (s ) =

cp

x
x

p
m

x
x

cp

cg

C yv C y C nv C nr C n

y s y n r s U (n y v nv y ) s ( y v + n r )s + (Unv + y v n r )
2

(5) (6) (7) (8)

F (s ) =

98700 s + 445s + 98700 253000 H r (s ) = 2 s + 710s + 253000 394800 H ay (s ) = 2 s + 890s + 394800

Yawing moment

s mC

yv

, where s m = x m / d , where s f = x f / d

500 + 30M + 200


s mC
y

where Pr (s ) is the body rate transfer function, Pay (s ) is

the acceleration transfer function, F (s ) is the fin servo dynamics with maximum fin angle of 0.3 rad, H r (s ) is the rate gyro dynamics, H ay (s ) is the lateral
accelerometer dynamics, respectively. The cascade control structure of lateral accelerator is shown in Fig. 2. The arrangement of block-wise of the control configuration is shown in Fig. 3. In Fig. 3, C (s ) is the compensator, D (s ) is pre-filter. K r is the gyro gain and l r is the accelerometer moment constant. The control aim is to ensure accurate tracking of required acceleration manoeuvres in the presence of uncertainties in the aerodynamic characteristics, disturbances (wind gusts) and sensor noises. The controller produces the inputs to servo-actuators which rotate the fins. The normal acceleration in each plane and

Figure 2. Cascade control block diagram of lateral acceleration.

Figure 3. 5-DOF control system block diagram of lateral acceleration.

197

from the missiles centre of gravity, 0.7 m aft of the nose. This produces measured acceleration a y that contains a
m

The membership functions of these fuzzy sets for e ( k )

(9) l a is the accelerometer moment arm from the centre of gravity the accelerometer will sense the normal acceleration of the missile centre of gravity.
m

component of angular acceleration of a y = ay + l a .d (r ) / d (s )

N represents negative, P positive , ZO approximately zero , S small, M medium and B big, respectively. Thus NM stands for negative-medium, PB for positive big, and so on. The fuzzy sets Ci and Di may be either Big or Small and are
and e (k ) are shown in Fig. 4. In Fig. 4, characterized by the membership function shown in Fig. 5. In this study, we use triangular membership function for e ( k ) , e (k ) , K ' p and K 'd and our membership functions are normal, consistent and complete to make the system more linear. Figure 6 shows the membership functions for . Figures 7 and 8 show the relation of antecedent variables versus the output variables of the fuzzy system. Also, the rule bases are indicated in tables III, IV and V. Once the fuzzy output variables are obtained, the PID controller parameters are calculated from the following equations

FUZZY SELF-TUNING GAIN SCHEDULED CONTROL DESIGN The transfer function of a PID controller has the following form: G c (s ) = K p + K i s + K d s (10) where K p , K i and K d are the proportional, integral and derivative gains, respectively. Another alternative and useful equivalent form of the PID controller is: (11) where Ti = K p K i and Td = K d K p . Ti and Td are known as integral and derivative time constant. The discrete-time equivalent form of the PID control, for digital implementation used in this paper is given as: n K u (k ) = K p e (k ) + K iT s e (i ) + d e (k ) Ts i =1 (12) where u ( k ) is the control signal, e ( k ) is the error between the reference and the process output, T s is the sampling period for the controller, and e ( k ) = e ( k ) e ( k 1) . The parameter of the PID controller can be manipulated to produce various response from a given process. To find optimum adjustment of a controller for a missile is too hard, because its inputs are changing all the time and we need a controller that can control the missile all the time. An online gain scheduling of the PID controller based on fuzzy rules is proposed in this study. It is assumed that K p and K d are in prescribed ranges [ K p min , K p max ] and [ K d min , K d max ]. K p and K d are normalized into the range between zero and one by the following linear transformation:

III.

Gc ( s ) = K p (1 + 1 (Ti s ) + Td s )

K p = ( K p max K p min ) K ' p + K p min


K d = ( K d max K d min ) K ' d + K d min
(17)

Ki = K p ( K d )
2

Here the fuzzifiers are singleton, defuzzifiers are centroid, t-norm and implication function are min-max.

Figure 4. Membership functions for antecedent variables.

Figure 5. Membership functions for consequent variables.

K ' p = (K p K p min ) (K p max K p min )


K
' d

(13)

(14) The integral time constant is determined with reference to the derivative time constant, i.e. T i = T d (15) and the integral gain is thus obtained by

= ( K d K d min ) ( K d max K d min )

Figure 6. Membership functions for

K i = K p (T d ) = K p
The fuzzy rule is:

( K d )

(16)

if e ( k ) is Ai and e (k ) is B i , then K ' p is C i , K 'd is D i and = i ( i =1,2,, m )

Ai , Bi , Ci , Di are fuzzy sets, i is a singleton fuzzy set.

198

Figure 7. Rule surface of

K ' p and K ' d .

IV. SIMULATION RESULTS In this section, the fuzzy self-tuning controller is applied on the missile to evaluate the performance and efficiency of the closed loop control system. Figure 9 shows the performance of the closed loop control using a classical PID. It shows a weak performance in the transient state with highly oscillatory behavior in controlling the lateral acceleration of the missile. Also steady state error exists in the behavior. Figure 10 shows the performance of the proposed fuzzy self-tuning controller to track the desired reference signal. Not only the oscillation has been almost removed in the transient behavior of the control system, but also the steady error has been omitted. Figure 11 shows a comparison between the performance of the classical PID controller versus fuzzy PID controller one. To evaluate the performance of the control system to reject the disturbance, a step disturbance is entered to the system at t=6 sec. The simulation result is shown in Figure 12. The result shows that the disturbance is completely attenuated in a few seconds after entering the disturbance.
1.4 1.2

Figure 8. Rule surface of

.
amplitude

0.8

Table III. Rule base for K ' p


e (k )

0.6

0.4

e(k )

NB NM NS ZO PS PM PB

NB B S S S S S B

NM B B S S S B B

NS B B B S B B B

ZO B B B B B B B

PS B B B S B B B

PM B B S S S B B

PB B S S S S S B
a p d m litu e

0.2

3 time

Figure 9. Step response for classical PID control.


1.4

1.2

0.8

Table IV. Rule base for K 'd .


e (k )

0.6

0.4

e(k )

amplitude

NB NM NS ZO PS PM PB

NB S B B B B B S

NM S B B B B B S

NS S S B B B S S

ZO S S S B S S S

PS S S B B B S S

PM S B B B B B S
. PM 2 3 3 4 3 3 2

PB S B B B B B S

0.2

3 time

Figure 10. Step response for fuzzy gain scheduled PID control.
1.4

1.2

0.8

0.6

Table V. Rule base for


e (k )

0.4

0.2

e(k )

NB NM NS ZO PS PM PB

NB 2 3 4 5 4 3 2

NM 2 3 3 4 3 3 2

NS 2 2 3 3 3 2 2

ZO 2 2 2 3 2 2 2

PS 2 2 3 3 3 2 2

PB 2 3 4 5 4 3 2

3 time

Figure 11. Step response comparison between the classical PID controller and fuzzy gain scheduled PID control.

199

1.5

0.5

5 time

10

Figure 12. Disturbance rejection performance of fuzzy gain scheduled PID control.

V. CONCLUSION REMARKS This paper has shown that fuzzy gain scheduling is an easy and efficient way to control a missile system with helping human knowledge and experience. The proposed gain scheduling scheme uses fuzzy rules and reasoning to determine the PID controller parameters. Good tracking and disturbance rejection were obtained in a wide range of inputs with minimum error. REFERENCES
[1] C. K. Lin and S. D. Wang, An adaptive controller design for bank-to-turn missile using ridge Gaussian neural network, IEEE trans. neural networks, vol.15, no.6, pp.1507-1516, November 2004.

R. Sutton and P. J. Craven, A fuzzy autopilot design approach that utilizes nonlinear consequent terms, Jr. Marine Science and technology, vol.9, no.2, pp.65-74, 2001. [3] E. J. Hughes, A. Tsourdos, and B. A. White, Multiobjective design of a fuzzy controller for a nonlinear missile autopilot, IEEE Intl. Symp. Comp. Aided Contr. Syst. Design Proc., pp.1520, Sept. 2002. [4] D.-W. Gu, Petkov and M. M. Konstantinov, Robust control design with matlab, Springer, Berlin, 2006. [5] T. Screenuch, A. Tsourdos, E. J. Hughes, and B. A. White, Fuzzy gain-scheduled missile autopilot design using evolutionary algorithms, IEEE Trans. Aerospace Electronics, vol. 42, no. 4, pp. 1323-1339, October 2006. [6] Z. Y. Zhao, M. Tomizuka, and S. Isaka, Fuzzy gain Scheduling of PIDcontrollers, IEEE Trans. on Systems, Man, Cybernetics, vol. 23, no. 5, pp. 1392-1398, September/October 1993. [7] W. J. Rugh, Analytical framework for gain scheduling, IEEE Contr. Mag., vol. 11, pp. 799-803,1993. [8] R. Sutton and P. J. Craven, A fuzzy autopilot design approach that utilizes nonlinear consequent terms, Jr. Marine Science and Technology, vol. 9, no. 2, pp. 65-74, 2001. [9] C. K. Lin and S. D. Wang, An adaptive controller design for bank-to-turn missile using ridge Gaussian neural network, IEEE Trans. Neural Networks, vol.15, no.6, pp. 1507-1516, November 2004. [10] G. M. Siouris, Missile guidance and control systems, Springer, New York, 2004.

[2]

a p d m litu e

200

Das könnte Ihnen auch gefallen