Fault-tolerant control algorithm of neural network based on Particle Swarm Optimization

A fault-tolerant control method combining fault diagnosis and fault-tolerant control is proposed for sensor faults. A BP neural network based on Particle Swarm Optimization algorithm is used to estimate system states and fault parameters of the constructed model for sensor faults. The estimated fault parameters are processed by the modified Bayes classification algorithm to achieve sensor faults diagnosis, separation and estimation on-line, and sensor faults are described as "equivalent bias" vectors to realize fault-tolerant control by compensation algorithm. Simulation results for continuous stirred tank reactor (CSTR) show good convergence of the approach and strong fault-tolerant ability for sensor faults.