Vous êtes sur la page 1sur 6

2010 IEEE Symposium on Industrial Electronics and Applications (ISIEA 2010), October 3-5, 2010, Penang, Malaysia

Pitch Control System Using LQR and
Fuzzy Logic Controller
Nurbaiti Wahid Mohd Fua’ad Rahmat
Faculty of Electrical Engineering Faculty of Electrical Engineering
Universiti Teknologi MARA Universiti Teknologi Malaysia
23000 Dungun, Terengganu, Malaysia 81310 Skudai, Johor, Malaysia
nurbaiti@tganu.uitm.edu.my fuaad@fke.utm.my

Abstract—This paper presents a comparative assessment Pitch is controlled by the rear part of the tail plane's
based on time response specification performance between horizontal stabilizer being hinged to create an elevator. By
modern and intelligent controller for a pitch control system moving the elevator control backwards the pilot moves the
of an aircraft system. The dynamic modeling of pitch elevator up (a position of negative camber) and the
control system is considered on the design an autopilot that downwards force on the horizontal tail is increased. The
controls the pitch angle of an aircraft. It begins with a angle of attack on the wings increased so the nose is
derivation of suitable mathematical model to describe the pitched up and lift is generally increased. In micro-lights
dynamics of an aircraft. To study the effectiveness of the and hang gliders the pitch action is reversed and the pitch
controllers, the Linear Quadratic Controller (LQR) and control system is much simpler, so when the pilot moves
Fuzzy Logic Controller (FLC) is developed for controlling the elevator control backwards it produces a nose-down
the pitch angle of an aircraft system. Simulation results for pitch and the angle of attack on the wing is reduced. The
the response of pitch controller are presented in time pitch angle of an aircraft is controlled by adjusting the
domain. Finally, the performances of pitch control systems angle and therefore the lift force of the rear elevator. The
are investigated and analyzed based on common criteria of aerodynamic forces (lift and drug) as well as the aircraft’s
step’s response in order to identify which control strategy inertia are taken into account. This is a third order,
delivers better performance with respect to the desired pitch nonlinear system which is linearized about the operating
angle and pitch rate. It is found from simulation, LQR point.
controller give the best performance compared to fuzzy This work presents investigation into the development
logic controller. of pitch control schemes for pitch angle and pitch rate of
an aircraft systems. Pitch control systems with full state
Keywords—Aircraft; Flight control; Autopilot; Longitudinal feedback controller are investigated. A modern controller
dynamic; LQR; Fuzzy logic. (LQR) and intelligent fuzzy logic controller (FLC) is
developed for control the pitch of an aircraft systems.
I. INTRODUCTION Performance of both control strategy with respect to the
Today’s aircraft designs rely heavily on automatic pitch angle and pitch rate is examined. Comparison of
control system to monitor and control many of aircraft’s both control schemes to the system performance of
subsystem. The development of automatic control system aircraft system is presented and discussed
has played an important role in the growth of civil and
military aviation. Modern aircraft include a variety of II. MODELLING OF A PITCH CONTROL
automatic control system that aids the flight crew in Flight control system has been designed using
navigation, flight management and augmenting the mathematical models of the aircraft linearized at various
stability characteristic of the airplane. For this situation an flight condition parameters varied with the flight
autopilot is designed that control the pitch of aircraft that operating conditions [9]. This work is developed to
can be used by the flight crew to lessen their workload control the pitch angle of an aircraft for pitch control in
during cruising and help them land their aircraft during order to stabilize the system when the airplane is nose up
adverse weather condition in the real situation [1]. The and nose down. The pitch control system considered in
autopilot is an element within the flight control system. It this work is shown in Fig. 1 where Xb, Yb and Zb represent
is a pilot relief mechanism that assists in maintaining an the aerodynamics force components. θ, Ф and δe represent
attitude, heading, altitude or flying to navigation or the orientation of aircraft (pitch angle), orientation of
landing references [2]. Designing an autopilot requires aircraft (roll angle) in the earth-axis system and elevator
control system theory background and knowledge of deflection angle.
stability derivatives at different altitudes and Mach
numbers for a given airplane [3]. Lot of works has been
done in the past to control the pitch of an aircraft for the
purpose of flight stability and yet this research still
remains an open issue in the present and future works [4],
[5], [6], [7] and [8].

978-1-4244-7647-3/10/$26.00 ©2010 IEEE 389

The equations are replaced by a variables or reference value plus a perturbation or disturbance. δ = δ o + Δδ The longitudinal stability derivatives parameter used are denoted in Table I. (4) Pitch Angle. QS c = 38596ft. r = ψCθ C Φ − θS Φ .8976 Pitching rate Xq = 0 Zq = 0 Mq = -2.369 Mu = 0 velocities Xw = 0. The parameter include in dimensional are determined. (5) roll Angle. Second.02 Mw = -0. q = θC Φ + ψCθ S Φ . and θ = qC Φ − rS Φ . QS = 6771lb. pitch and yaw system must be model in terms of mathematical equations. TABLE I. v. r represent the angular rates about roll. 2. Equation (1). u = uo + Δu v = vo + Δv w = wo + Δw p = po + Δp q = qo + Δq r = ro + Δr X = X o + ΔX M = M o + MY Z = Zo + ΔZ Figure 2.05 Elevator Xδe = 0 Zδe = -28. Definition of force. First.8lb/ft2. 390 . The aerodynamics moment components for roll.016s is considered. LONGITUDINAL STABILITY DERIVATIVE PARAMETERS Components Longitudinal Derivatives X-Force Z-Force Pitching (S-1) (F-1) Moment (FT-1) Rolling Xu = -0. In order to analyze pitch and yaw axis are represent as L. M and N. The term the dynamics of aircraft in flight.045 Zu = -0. modeling of the pitch control system.8 Angle of attack X α = 0 Z α = 0 M α = -0.05 Yawing velocities X w = 0 Z w = 0 M w = -0. w represent the velocity components Mathematical modeling involves process to describe of roll. 2 shows the forces. (2) and (3) should be linearized using small disturbance theory. the aircraft dynamic p. the aircraft is steady state cruise at constant altitude and velocity. as shown below. components in the body fixed coordinate of aircraft The section provides a brief description on the system. p = Φ −ψSθ . axis while term u.15 Mδe = -11. Description of pitch control system.036 Zw = -2. X − mgS θ = m(u + qv − rv ) (1) Z + mgCθ CΦ = m( w + pv − qu) (2) 2 2 M = I y q + rq ( I x − I z ) + I xz ( p − r ) (3) It is required to completely solved the aircraft problem with considering the following assumption: (1) rolling rate.  and β are represents as the dynamics of a system in a set of differential equations. the change in pitch angle does not change the speed of an aircraft under any Fig.lb and ( c / 2u0 ) = 0. derivatives Q = 36. angle of attack and sideslip. balance out each other.874 deflection A few assumption need to be considered before continuing with the modeling process. (2) yawing rate.051 Xα = 0 Z α = -355. pitch and yaw axis. Φ Φ θ Φ θ (6) Yaw Angle. the following General Aviation Airplane [1] is used in system analysis dynamic equations include force and moment equations and modeling. ψ = (qS Φ + rCΦ ) sec θ .42 M α = -8.  = p + qS T + rC T . (3) pitching rate. thus the thrust and drag are cancel out and the lift and weight Figure 1. In this study the data from Referring to the Fig. moments and velocity in body fixed coordinate. moments and velocity circumstance. 1 and Fig. q.

This is done by in pitch rates to the change in elevator angle in the choosing two parameter values.02 1 0⎤ ⎡ Δα ⎤ ⎡ 0. settling time less than 5 second. Δδe (s) s − (Mq + Mα + Zα / u0 )s + (Zα Mq / u0 − Mα ) In designing LQR controller. With a full-state feedback controller Δ δ e ( s ) s 3 + 4.9868 − 2. by tuning the value of x = 500. Nbar can be found using the user- defined function which can be used in m-file code. The method used in simulation work is done by exported both ⎡ Δα ⎤ ⎡ − 2. 3. Nbar are found to be. By manipulating the (4). (6) and substituting the parameters values of the longitudinal stability derivatives.9476 0⎥ ⎢ Δq ⎥ + ⎢11.2 ⎛d ⎞ (4) radian (11. Nbar = 22. Furthermore. the reference flight condition is (LQR) and Fuzzy Logic Controller (FLC).3607] = ⋅ (11) Δδe (s) s s2 − (M q + Mα + Zα / u0 )s + (Zα M q / u0 − Mα ) In order to reduce steady state error of the system output. the performance of both control strategies.0 0 0. Q.941s all the states are feedback.360. percentage of overshot less than 10% and steady state error less than 2% for controlling the pitch angle of 0. In this work.7304 s + 22 . 391 . v0 = p0 = q0 = r0 = Φ 0 = ψ 0 = w0 = 0 .6929 22. If x is Therefore the transfer function of the pitch control increased even higher. Δq ( s ) = sΔ θ ( s ) (9) R = 1. multiply that by the chosen The transfer function can be represented in state-space gain K. input R = 1 and Q = following way. The steady-state value of the states should be computed. After linearization four considerations have to be met which are rising time the (4). computing the input. the following values of matrix K are obtained. ⎜ − X u ⎟ Δu − X w Δw + ( g cos θ o )Δθ = Xδ e Δδ e ⎝ dt ⎠ A. Block diagram of the system with matrix. 4. R]. less than 3 second.9676 s 2 + 12 . CT*C where CT is the matrix transpose of C from state equation (14). Full-state feedback controller with reference input. Δq(s) − (Mδe + Mα Zδe / u0 )s − (Mα Zδe / u0 − MδeZα / u0 ) (7) = 2 Figure 3. = ⋅ Δδ e (s) s Δθ (s) Consequently. lqr function in Matlab The transfer function of the change in pitch angle to the can be used to determine the value of the vector K which change in elevator angle can be obtained from the change determined the feedback control law.16 ⎤ value of matrix K and constant gain. K and gain Nbar. Q = [0 0 0. The system can be stabilized using ⎝ dt ⎠ ⎝ dt dt ⎠ full-state feedback system. METHODOLOGY In this section. a assumed to be symmetric and the propulsive forces are few of design specification have to be set to investigate assumed to remain constant. a value of constant gain Nbar should be added Δθ ( s ) 11 . The configuration of this control system is shown in Fig.0 0 x].5 degree). the values of x = 500 is chosen because it satisfied the design requirements while keep x as small as possible. and used a new value as the reference for form and output equation as state by (13) and (14). two control schemes are proposed and describe in detail which is Linear Quadratic Regulator Figure 4. Nbar as shown in ⎢ Δq ⎥ = ⎢− 6. the following transfer function for the change in the pitch rate to the change in elevator deflection angle is shown as (7) obtained. The controller can be tuned by changing the nonzero elements in q matrix which is done in m-file Δq = Δθ (8) code as obtained.578 (12) = after the reference. (5) and (6) are obtained. LQR Controller ⎡ d ⎤ ⎡ d ⎤ (5) − Zu Δu + ⎢(1 − Z w ) − Z w ⎥Δw − ⎢(uo + Z q ) − g sinθ0 ⎥Δθ = Zδe Δδ e LQR is a method in modern control theory that used ⎣ dt ⎦ ⎣ dt ⎦ state-space approach to analyze such a system. Δθ (s) 1 Δq (s) (10) K = lqr [A. Δθ (s) 1 −(Mδe + Mα Zδe / u0 )s − (Mα Zδe / u0 − MδeZα / u0 ) Matrix K = [-0. Using ⎛ d ⎞ ⎛ d2 d⎞ state space methods it is relatively simple to work with a − M u Δu − ⎜ M w + M w ⎟Δw + ⎜⎜ 2 − M q ⎟⎟Δθ = M δe Δδ e (6) multi-output system. For this controller design. the value of constant  gain.5704 1. B. (5). ⎣⎢ Δθ ⎦⎥ ⎣⎢ 0 1 0⎦⎥ ⎣⎢ Δθ ⎦⎥ ⎢⎣ 0 ⎦⎥ ⎡Δα ⎤ y = [0 0 1]⎢⎢ Δq ⎥⎥ + [0] (14) ⎢⎣ Δθ ⎥⎦ III. But for this case. This implies that.7304⎥[Δ ] (13) ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ δe Fig. For convenience. improvement to the response system is obtained in (11) and (12) respectively. should be obtained even more.

the error (e) is and the output ‘output1’. The change of error (Δe) is generated by the derivation of the error. functions. and implementing the expert knowledge in a form of IF- THEN rule structure. This response is transform into fuzzy rules using the formula obtained below. Fuzzy set of the input ‘∆error’ generated from the under damped response curve. the triangular and trapezoidal membership functions are chosen for each fuzzy set. 6. 5 shows the overall closed-loop system for FLC triangular so each membership function has three with the pitch control of an aircraft. which is negative (N). FLC inputs and one output). A two input. Fuzzification involves the conversion of the input and output signal into a number of fuzzy represented values (fuzzy set). In designing FLC. the standard fuzzy rules Figure 7. 5. complex 4 Z N N mathematical computation and real-time computation 5 Z Z Z need. TABLE II. Fuzzy Logic Controller RULES FOR THE FUZZY CONTROLLER In this work. e(k ) = r (k ) − y (k ) and Δe(k ) = e(k ) − e(k − 1) In this work. It 8 P Z P uses an imprecise but very descriptive language to deal 9 P P P with input data more like a human operator. the fuzzy controller for pitch approach to solving control problem rather than control of an aircraft has a total of nine membership attempting to model a system mathematically. in Table II. fuzzy logic controller has been applied for Error. The appropriate membership function to represent each fuzzy set need to be defined and each fuzzy set must have the appropriate universe of discourse. Figure 6. FLC is conceived 1 N N N as a better method for sorting and handling data but has 2 N Z N proven to be an excellent choice for many control system 3 N P N applications because of non-linearity. the triangular and trap membership system performance and the rate at which the error function are chosen for each fuzzy set. Each membership function is constrained to be Fig. de Delta u. Fuzzy set of the input ‘outout1’. Fig. These are nine rules that have been utilized in designing the controller and the rule is defined Figure 8. It is very robust and forgiving of operator and data input and often works when first implemented with little or no tuning. It can be built into anything from small. Fig. computed by comparing the reference point (desired angle) with the plant output.4 to 0. zero (Z) and positive (P). e Delta error. 8 shows the fuzzy set of the input ‘error’ and ‘Δerror’ control signal (Δu). fuzzy controller are the error (e) which measures the In this study. In addition. 392 . Figure 5. the membership functions are evenly distributed so that the tuning process of the controller can be easily done. 7 and changes (Δe). The error and change of error is fed to the fuzzy controller through a multiplexer. The universe of discourse is set between -0. Fuzzy set of the input ‘error’. whereas the output is the change of the Fig.4 that implies the range of pitch angle (±0. fuzzy logic controller plays the Since there are a total of three fuzzy variables (two best to fit the requirements in such cases. The inputs to the parameters (a modal point and two half-width). Based on these properties.4 radian). one output fuzzy pitch control can be designed by defining error as the reference angle minus the measured angle. hand-held 6 Z P P 7 P N P products to large computerized process control systems. Fuzzy logic controller in feedback loop of pitch control system. du stabilization of the pitch control system. Each fuzzy set consists of three types membership function. From the Fig. and each fuzzy variable has three incorporates a simple rule-based “If X and Y then Z” membership functions. B.

5 0. the results clearly demonstrate that LQR controller has the fastest response with the settling time of 0. 12. This can be indicating that LQR controller can has the best performance as compared to fuzzy logic handle the effect of disturbances in the system. from the result The fuzzy logic controller provides good performance in obtained in Fig. 12. This controller is able to give a good response without produce any overshoot. for (desired angle) with the plant output and the change of the percent of overshoot (%OS) fuzzy logic controller have error (Δe) which generated by the derivation of the error.05 0 1 2 3 4 5 6 7 8 pitch angle between LQR and fuzzy logic controller is Time (sec) shown in Table III quantitatively. 0.15 (ess) (%) Pitch angle (radian) 0.05 0 0 1 2 3 4 5 6 7 8 Time (sec) Figure 11.5%.03 second. 0.35 0 (%OS) 0. the the percent of overshoot (%OS). reference value respectively.15 A unit step command is required in order for pitch angle Pitch angle (radian) to follow the reference value of 0.2 radian = 11. By referring to the Fig. LQR has 4.1335s 1.05 The system response with the linear quadratic regulator 0 (LQR) is shown in Fig.1 The pitch control system with both LQR and fuzzy logic controller produced the response of pitch angle. SUMMARY OF PERFORMANCE CHARACTERISTIC FOR PITCH ANGLE (Ts) about 2.2 In this section.01%. The response is comparatively fast that give the settling time TABLE III. Pitch angle response with LQR and fuzzy logic controller. However.1335 second. The results clearly shows that LQR controller is 0. 10 shows the closed loop system response of the settling time (Ts) that is 0. θ with fuzzy logic controller.01 0. Pitch angle response with fuzzy logic controller. For For comparison of controller performance.25 Reference Pitch angle-LQR Percent Overshoot 4.002s (Ts) 0.15 Figure 9. 9. 0. 393 .05 Reference 0 1 2 3 4 5 6 LQR Time (sec) FLC 0.1 0. it can be concluded that term of percent overshoot that is 0%. 11 and Fig. The summary for the performance characteristics of the step response for the -0. the proposed of control schemes are implemented and the corresponding results are presented.25 Reference IV. the best range which is 0%.1335 second.25 0 0. controller in term of rising time (Tr) that is 0.2 -0. The results also show have been applied to fuzzy logic controller which is the the LQR controller is good in eliminating the error from the error (e) that computed by comparing the reference point system which is almost tend to the zero value.03s (Tr) Settling Time 0.05 0.002 second and rise time (Tr) about 1. Pitch angle (rad) 0. it can be observed that the pitch angle follows the the pitch angle as compared to the fuzzy logic controller. 0.1 0. 11 and the bar graphs small steady state error (Ess) and it is within the limit that in Fig.2 Steady-state Error 0. Rising Time 0.01%. 10. RESULTS AND DISCUSSION Pitch angle-FLC 0. Two inputs steady state error (Ess) about 0.3655s 2. Therefore. θ. Pitch angle response with LQR controller.3655 second and rising time of 0.5 degree. As depicted from the LQR controller provide higher ability in controlling Fig. Fig. Figure 10.35% which response for pitch control of an aircraft system using is met the desired requirement of controller design. 9 and Table III. LQR and fuzzy logic controller are shown with overall Furthermore the LQR controller tends to produce very response of both controllers in Fig. The results also demonstrated that the artificial Response Pitch angle fuzzy logic controller can eliminate the effect of characteristic LQR Fuzzy Logic disturbances in the system up to 0.3655 second and percent of pitch angle.

following by develop more advanced and robustness control techniques. A New Paradigm for an Old Art”. Stojiljkovic. Simulation and analysis results show that. the control approach of LQR and fuzzy logic is capable of controlling the pitch angle of the aircraft system for value of 0. Engineering and Technology. [4] Pavle Boskoski. and Steven R. No 5. [7] D. Eick. “Pitch Control of the Space Shuttle Training Aircraft”. 2001. Egypt. 2005. Vasov. LQR and fuzzy logic are successfully designed and presented. CONCLUSION The validated model of pitch control of an aircraft is very helpful in developing the control strategy for actual system. Nelson. Sujit Saraf. C. Berenji. Based on the result and the analysis. Postlethwaite. Cvetkovic. This can be achieved by reducing the error signal which is the difference between the output angle the desired angle. 1998. Part I. Vol 215. “Auto Landing Using Fuzzy Logic”. Ping-Wei Chang. 394 . the proposed control algorithm can be implements to real plant for validating of theoretical result. Cairo. REFERENCES [1] R. L. Lee. “Quantifying the Robustness of Flight Control Systems Using Nichols Exclusion Regions and the Structured Singular Value”. 6th International PhD Workshop on Systems and Control. “Nonlinear Pitch Autopilot Design with Local Lines Linear System Analysis”. LQR controller relatively give the better performance compared to fuzzy logic controller in controlling the pitch angle of an aircraft system. No 3. Y. Journal of Mechanical Engineering. effort can be devoted through adding another element that make up the control system. V. “The Application of the Root Locus Method for the Design of Pitch Controller of an F-104A Aircraft”. Slovenia. 2003. Stojche Deskoski. Beside. G. Proc InstnMech Engrs. a conclusion has been made that.2 radian (11. [8] Hamid R. Figure 12. A Reconfiguration Scheme for Flight Control Adaptation to Fixed Position Actuator Failures. 2001. Vol 5. 2001. Swanson. “Sliding Mode Control of Pitch Rate of an F-16 Aircraft”. Second Edition. [2] Thomas J. International Journal on Applied Science. Redling. University of Florida. [6] Robert S. Sridhar Seshagiri. Vol 9. McGraw Hill. IEEE Transaction on Control System Technology. [3] B.C. R. Vol 55. 2009. International Conference on Automatic Control and Systems Engineering. Cho.5degree). S. Choe. Bates. “Integrated Flight Control System. Biljana Mileva. 2005. Pitch control of an aircraft is a system which requires a pitch controller to maintain the angle at it desired value. Two controllers. Ph. IEEE Aerospace and Electronic Systems Society (AESS) Systems Magazine. Mitrovic. D. [9] Ekprasit Promtun. D Theses. Flight Stability and Automatic Control. For further research. Performance comparison between LQR and fuzzy logic controller. [5] D. Kureemun and I.