Design of variable structure fault tolerant controller for nonlinear system

A design method of variable structure fault tolerant controller (VSFTC) is presented for a class of systems with unknown structure. It is inserted into a normal system to substitute for the primary controller automatically when faults occur. RBF neural network is used to identify system and fault detection is realized by threshold test approach. The system can run in three modes: common controller is in use when it is normal, otherwise, changes into adaptive neural network controller when step fault occurs or changes into prediction controller when excursion fault occurs. The simulation for continuous stirred tank reactor (CSTR) shows that the approach presented can guarantee system´s performance and reduce the effects of faults.