Sonar array servo system based on diagonal recurrent neural network

Abstract

The sonar array servo system based on traditional PID can not perform well when the parameter changes. So for improvement, the sonar array azimuth servo system controller is designed based on both the diagonal recurrent neural network (DRNN) and the traditional PID in this paper. In fact, the DRNN controller serves as a monitor of the PID controller. In the initial period, the DRNN controller is trained by the output of the PID controller. The DRNN controller begins to act effectively when the output of the PID controller trends towards 0. The DRNN controller takes the place of the traditional PID controller step by step. This paper also selected suitable parameter and carried on systematic simulation. The simulation results show that compared with the traditional PID controller, the controller based on DRNN has lower tracking errors, shorter settling time and a stronger robustness.