Active Fault Tolerant Networked Control System Based On Neural Network Control Technology

Networked control system(NCS) is the system that feedback control loop communicate through a real-time network. Focusing on the situation of the existence of time delay, external disturbances and the actuator failures, the problem of active fault tolerant control of NCS is inverstigated in this dissertation. The main work are as follows:Firstly, considering the external interference as well as actuator failures in a short time Single Input Single Output NCS, the neural network fault tolerant strategy based on model reference adaptive control is presented, using the tracking error to design neural network output weight learning algorithm, and then prove the closed-loop system is uniformly ultimately bounded stable based on Lyapunov stability theory. The remote servo control system simulation platform is designed based on TrueTime toolbox of Matlab/Simulink, the simulation results show the effctiveness and feasibility of this method.Secondly, the random short time delay Multiple input Multiple output NCS without considering the data dropout is modeled as random switching system based on the estimated delay by Average Delay Window technology. Based on switching system theory and Lyapunov stability theory, the adaptive RBF neural network controller is designed to compensate the impact of actuator failure and time delay, ensure NCS with actuator failure is stable and track the reference model. The simulations are given to illustrate the feasibility of the method.Finally, with respect to the short time delay NCS without considering data dropout, the network just exist between sensor and controller. The adaptive neural network fault diagnosis observer is designed to estimate the system state and actuator failures.The active fault tolerant controller is designed based on Lyapunov stability. The closed-loop system is uniformly ultimately bouned stable. With respect to the different types of actuator failures, the remote servo control system simulation platform is designed based on TrueTime toolbox, simulation results show the effectiveness of this method.