Robust Fault Diagnosis For The Flight Control System Of A Jumbo Jet Transport Aircraft

The program of developing large civil aircrafts was established as oneof the long-term science and technology development plans in China. Toreach the world advanced level, China makes emphasis on independentdevelopment of safe and reliable flight control systems. But the faultoccurred in the safty-critical flight control system would lead to seriousconsequences. Supported by Chinese Civil Aircraft Development Plan, themain purpose of this thesis is to study robust FD (Fault Diagnosis) filtertheory with application into diagnosing control surface faults of a jumbojet transport aircraft, and the disturbance in measured signal and parametervarying effect in the mathematical model are also considered. The studiesare extended progressively from model establishing, robust diagnosis filterdesign methods to robust diagnosis system design.1. The theory of mathematical modeling for a jumbo jet transportaircraft is studied, including describing the6-DOF rigid aircraft dynamicsusing nonlinear differential equations; and calculating aerodynamic forceand torque via aerodynamic coefficients together with flight states andcontrol surface angles; and getting the linearized LTI (Linear TimeInvariant) models via small disturbance theory; and an affine LPV (LinearParameter Varying) model for the aircraft via affine transformation.Meanwhile a total mathematical model of the B747-100/200aircraft isestablished, and a software package of dynamics simulation forB747-100/200is developed based MATLAB and SIMULINK.2. The problem of designing robust fault diagnosis filter withparameter-varying mixed H∞/H–index is studied for the control surface faults of a jumbo jet transport aircraft. By using LFT (Linear FractionalTransformation) method, the parameter-varying H∞/H–mixed index istransformed into a parameter-varying H∞index using the Loop-Shapingtheory, then the aboved parameter-varying H∞problem is transformed intoa H∞robust performance problem for the LTI system with structureduncertainty through the LFT tool, and the sufficient conditions of theexistence of the diagnosis filter are achieved via the scaled bounded reallemma, and the parameters of the filter are obtained by MATLAB SDP(Semi-Definition Programming) toolbox. This LPV filter has followingadvantages:(1)the infinite dimensions problem for the linearparameter-varying filter is simplified as a LTI problem via LFT, and thecalculating level is much less than polytopic-and grid-methods;(2) theproposed filter has the same varying parameter structure as the monitoredplant, and it has adaptive capability for the varying parameters.(3) theperformance requirement is easy implemented only through adjustingweighting function, and interconnection struction of the flight controlsystem in the proposed algorithm and the solutions of the assoicated linearmatrix inequalities are able to obtain via Matlab toolbox, and then it hasgood engineering value.3. The proposed robust fault diagnosis filter is applied into detectingthe injuried faults of elevators, ailerons and rudders of the B747-100/200aircraft. Considering the decoupled effect of longitudinal and lateraldynamic, elevator FD filter and aileron-rudder FD fitlter was separatelydesigned, and a diagonal weight function was used to achieve isolation ofaileron and rudder fault. Two LTI model based robust FD filters weredesigned, and two affine LPV model based robust FD filters were designed.Through the analysis of simulation for fault and FD filter response, someconclusions were made,(1) the elevator fault occurred in steady flightcondition can not be detected;(2) the LTI model based robust FD filtercannot fulfill the performance proposed in local condition when flightparameters varied largely, but the LPV model based robust FD filter, though there exists some degree of descent in performace relatived to LTIfilter, achieves anticipant performance through all parameter space;(3)aileron-rudder fault residual can achieve larger bandwidth than elevatorfault residual.In general, this thesis studies a LPV robust FD filter design method,and applies this method into a jumbo jet transport aircraft of B747to faultdiagnosis of control surfaces. Simulation results show this algorithm notonly has good performance in detect diagnosis, but also has excellentcapbability of robust adjustness for varying flight parameters.