Applying RBF neural network to missile control system parameter optimization

Abstract

The PID (proportional, integral, differential) control method is widely applied to missile attitude control. The usual empirical method for optimizing the three control parameters of Kp, Ki and Kd can not optimize them on line and in real time. The paper presents the PID parameter optimization method that uses RBF neural network, applies it to a missile's longitudinal control system parameter optimization and verifies its effectiveness through numerical simulation. The simulation results demonstrate preliminarily that the use of RBF neural network can optimize the missile control system parameters on line and in real time.