Uncertain parameters variable structure neural network identifier in spherical actuator control system

Owing to spherical actuator control system uncertain parameters generated control errors, our project established offline learning neural network identifier to correct systematic errors. But BP neural network identifier only learning or training offline, which can not be used for errors correcting online, that means control accuracy is susceptible to interference field. Therefore, this paper proposes a variable structure BP (VSBP) neural network identification algorithm to meet the needs of the online error correction control system. The simulation results show that VSBP neural network identifier solved the problem that BP neural network identifier can not meet the needs of the online error correction control system, improved noise immunity and reliability of the spherical actuator control system.