Particle Swarm Optimized Critic Neural Network Based Fault Tolerant Control For Unknown Nonlinear Systems

Modern industrial production equipments become more and more complex to satisfy requirements of production quality and economic efficiency.Meanwhile,these complex systems have always unknown dynamics,high degree of nonlinearities and uncertainties,which make it difficult to develop controllers.Moreover,the occurrence of faults is inevitable due to increasing components.Adaptive dynamic programming(ADP)is a near-optimal method emerging in the field of intelligent control.It has capabilities of selflearning and optimization,and widely applied in the fields of motion control,power systems and chemical processes.Note that the ADP-based fault tolerant control(FTC)strategy can guarantee both fault tolerant abilities and system optimality by integrating the ADP algorithm into the fault tolerant controller.This paper studies ADP theory and its application in FTC.The main contributions of this thesis are as follows:1.In this thesis,we study the ADP-based FTC for partially unknown continuous-time(CT)nonlinear systems.Firstly,the neural network identifier is developed to learn the unknown system dynamics,and a fault observer is designed for real-time estimation of actuator faults.Secondly,in order to eliminate the influence of actuator faults,a cost function that reflects the actuator fault is designed to tra nsform the FTC problem into the optimal control problem.Then,the particle swarm optimized critic neural network(PSOCNN)is developed to solve the Hamilton-Jacobi-Bellman equation(HJBE),which has a higher success rate in solving the HJBE compared to the gradient-based critic neural network(GDCNN).Finally,a simulation is given to verify the effectiveness of the proposed algorithm.2.In this thesis,a data-based(FTC)scheme is investigated for completely unknown CT nonlinear systems with actuator faults.Firstly,a neural network identifier based on particle swarm optimization(PSO)is constructed to model the unknown system dynamics.By utilizing the estimated system states,the PSOCNN is employed to solve the HJBE more efficiently.Then,a data-based FTC scheme,which consists of the NN identifier and the fault compensator,is proposed to achieve actuator fault tolerance.Based on Lyapunov stability theorem,the identification error and fault estimation error is guaranteed to be uniformly ultimately bounded(UUB).Finally,simulations are provided to demonstrate the effectiveness of the developed method.