Study On Fault-tolerant Control Algorithms For Deep Space Probe Under Actuator Failure

Deep space probe does not have on-line repair ability as it is far away from the earth.Fault from system or components may greatly affect the success of the exploration mission.In order to improve the safety and reliability of the mission,some fault tolerant control algorithms are studied systematically considering the actuator failure as the application background in this paper.The main research work is as follows:First,on-line actuator fault diagnosis algorithm is presented.Attitude control system itself is a nonlinear system,coupled with the environmental uncertainty and modeling error,it becomes into a more complicated nonlinear system,so a new fault detection and diagnosis algorithm based on BP(Back Propagation)neural network is used to diagnose the abrupt actuators' fault directly.On the other hand,Super-Twisting detection based second-order sliding mode observer is designed to diagnose the abrupt and slowly-varying actuators' fault.Second,an adaptive iterative learning algorithm based fault-tolerant control method for probe orbiting is designed.The asteroid orbiting probe suffers the repetitive disturbance of irregular asteroid gravity and solar light pressure,etc.Considering the uncertainties mentioned above,the fault-tolerant controller is designed to ensure the success of orbiting observation and direction missions,while maintain high precision attitude tracking requirements.Then,an active fault tolerant control algorithm based on improved Structure Adaptive Model Inversion(SAMI)is designed for Mars entry system.Passive fault-tolerant controller only works for the given fault,so active fault-tolerant controller is designed instead.Based on the previous BP neural network fault detection and diagnosis method,the improved fault tolerant control algorithm with RBF(Radial Basis Function)neural network is used to adjust the parameters of the controller on line,so that the fault attitude system has good dynamic performance.Lastly,a direct model reference adaptive control method without fault detection is introduced for Mars entry.The attitude control system should guarantee the dynamic performance in some case without knowing the exact fault information,but passive fault-tolerant control or active fault-tolerant needs priori knowledge of the fault.The designed algorithm combines the basic controller based on model performance index and the adaptive model reference controller,ensures a good tracking performance in the case of slight actuators' fault.