Study On Iterative-learning-strategy-based Fault Estimation And Fault-tolerant Control For Vehicle Formation Driving System

Vehicle formation driving system(VFDS)plays an important role in “intelligent traffic”,its performance highly depends on actuators and sensors with high accuracy.However,actuator fault and sensor fault frequently occur under long-term high load operation.Actuator fault and sensor fault lead to deterioration of VFDS performance or even serious traffic accident.To ensure the stability and safety of VFDS,fault diagnosis and fault-tolerant control are necessary.In this thesis,a class of VFDS which executes carrying tasks repetitively is considered.The actuator fault and sensor fault are analyzed,fault estimation and fault-tolerant control strategy are investigated and fault-reject method based on iterative-learning strategy is proposed.The main content and contributions are shown as follows:(1)The problems of VFDS about driving control and controller parameters tuning are investigated.Distributed PID controllers are designed according to performance requirements of VFDS.The stability of single vehicle and string stability of VFDS are proved by classical control theory.For PID controller parameters tuning,an improved particle swarm optimization(PSO)algorithm is designed which can search PID controller parameters quickly and automatically in reasonable space.With the help of improved PSO,the performance of PID controller is improved and the VFDS can satisfy the preset requirement under a variety of working conditions.(2)The actuator fault rejection problem of VFDS which executes carrying tasks repetitively is studied.The actuator fault estimator is designed based on iterative-learning-strategy.Convergence of proposed fault estimator is proved theoretically and feasible range of matrix parameters is also solved.Accurate fault estimation result is achieved within limited iterations.Distributed fault-tolerant controllers are designed in compensation control form based on fault estimation results.Considering the influence of high-frequency state noise on the results of actuator fault estimation,filter is introduced to process the fault estimation signal of actuator.(3)The sensor fault rejection problem of VFDS is discussed.The situation that both actuator fault and sensor fault occur during the driving of VFDS is considered.To avoid the influence of actuator fault,an augmented state observer is designed and the sensor fault signal can be separated from the observed result.By compensating the measured output signal with the estimated value of the sensor fault,the PID controller and the iterative-learning-strategy based actuator fault estimator are restored.Lyapunov’s stability theory is used to prove the stability of proposed sensor fault estimation algorithm,the feasible range of parameter matrix and preconditions are also solved.