Surface failure detection and evaluation of control law for reconfiguration of flight control system

Abstract

The problem of detecting actuator surface failures and evaluating the control law for reconfiguration of flight control systems is considered. The method used for detecting failures is based on tracking error criteria which are developed by considering linear combinations of measured residual tracking errors. The tracking errors are generated using a model of the aircraft multivariable dynamics. Measurement disturbances1 noise are also included and avoidance of false alarms is discussed. The detection and control law, which consists of a control mixer and a compensating signal, evaluation method is demonstrated using a six degree of freedom A-7D Digitac II aircraft model at 0.6 Mach and 15,000 ft. altitude. The simulation results indicate that failure detection can be fast and successful, and false alarms can be avoided by properly modifying the tracking error criteria. The method developed is independent of the feedback control input and therefore the controller of the system can be updated at different failure situations with absolutely no effect on the detection decision making. Aircraft damage in battles or hardware failure may destroy the stability properties of the flight control system and lead to the loss of an aircraft. Even though the control laws can be designed with a certain degree of robustness with respect to uncertainties which may include modeling errors due to failures, such errors have to be small enough for stability and acceptable performance. When a primary surface becomes inoperative due to combat damage or mechanical failure, the control laws may no longer effectively exploit the resulting control power and therefore may lead to instability. One way to improve safety-of-flight reliability is to add redundant hardware. The extra hardware, however, not only adds additional cost but also increases maintenance activity. A more efficient way to deal with reliability in case of surface failures is to use the redundant control effectors, which are present (for performance reasons) in most newer aircraft, to distribute the forces and moments of the failed surfaces to the remaining healthy control surfaces. Thus the objective of such reconfigaration is to take advantage of the existing redundancy and produce the forces and moments acting on the aircraft without requiring the addition of redundant hardware. This approach will increase the survivability of the flight control system and at the same time reduce support requirements. The purpose of this paper is to develop a system impairment detection (SID) scheme which will detect and *~epar tment of Electrical Engineering, University of Southern California. t* Lockheed Aeronautical System Company, Member AIAA. classify the type of surface failures and a control reconfiguration (CR) scheme to distribute the control power to the healthy control surfaces. The SID scheme consists of a detection error model (DEM) and a failure classification model (FCM). The DEM is developed by using an observer type model for the aircraft dynamics to generate error signals and a similarity transformation is used so that each transformed error signal is affected by only one control surface. By monitoring each error signal we can detect individual surface failures and proceed by inputing this information to the FCM which will decide about the type of failure, i.e., whether the surface is partially missing, stuck or floating. The CR scheme is based on the control mixer approach [1,2,3] and the use of a compensating input signal to accommodate the surface failures. The compensating signal is needed when the impaired surface leads to a non-zero input to the aircraft. The error criteria are modified in the presence of noise in order to avoid false alarms. The higher the noise level, the longer the detection time of the failure. Our methodology is applied to a six degree of freedom A-7D Digitac II aircraft model at a cruise configuration of 0.6 Mach, angle of attack at 4.2O and flying at 15 K feet [4]. The simulation results presented in section 4 demonstrate the effectiveness of the proposed approach. In all cases surface failures were accommodated successfully. In this paper we assume that the parameters of the aircraft model are fixed and known. The more realistic situation where the parameters are time-varying and unknown [5,6] is presently under investigation. 2. Failure Detection Scheme The block diagram of the failure detection scheme and control reconfiguration scheme is shown below: SID scheme CR scheme I Compensating Signal d 1 1