Neural Network Observer-based Fault Diagnosis And Tolerant Control For Flight Control Systems

As a kind of typical complex engineering systems,the control system of the aircraft has such characteristics as nonlinearity,dynamic instability and strong coupling.The aircraft suffers from external disturbances,actuator faults and sensor faults during flight which will inevitably affect the flight performance and safety.Thus,it is of great significance to fault diagnosis and fault-tolerant control of flight control system.Based on the previous research,this paper takes the flight control system of unmanned aerial vehicle(UAV)as the research object.Based on the observer theory,the fault diagnosis and fault-tolerant control method are designed.Specific research work is as follows:Aiming at the longitudinal flight control system of a fixed-wing UAV with actuator faults,the modeling uncertainty and external disturbances are considered.Firstly,based on the neural network technology and H_?technology,a fault diagnosis method based on neural network observer is put forward to effectively estimate the fault.Secondly,a fault tolerant control algorithm is designed based on static output feedback to compensate the impact of the fault.A fault diagnosis algorithm based on improved neural network observer is proposed for the nonlinear flight control system of quadrotor subject to actuator faults and external disturbances.The neural network fault estimation algorithm is improved,in which the weight and the center value can be updated online to avoid the difficulty of parameter selection.At the same time,the nonlinear term of the control system is effectively processed.Secondly,a fault tolerant controller based on neural network is designed which effectively restores the flight performance of the system.Furthermore,the fault diagnosis of quadrotor with simultaneous actuator faults and sensor faults is considered.Actuator and sensor faults are separated based on the system decoupling.A fault estimation algorithm based on improved neural network observer is proposed for the subsystem with actuator faults.The disturbance and sensor fault in another subsystem are expanded to be states,a fault diagnosis method based on sliding mode observer is proposed to ensure the robustness of the system in the presence of disturbance.Finally,the fault diagnosis and fault-tolerant control method based on improved neural network observer proposed in this paper is applied to 3-DOF flight simulation platform.The experimental results show that the proposed method can effectively diagnose the fault,and can repair the system performance.