Improved Diagonal Recurrent Neural Network Approach to Tracking Control of Servo System

The improved Diagonal Recurrent Neural Network (DRNN)structure was proposed through combining of the DRNN and the Radial Basis Function(RBF). And the composite tracking controller was designed based on the improved DRNN for a dynamic servo system. The controller is composed of identifier,feed forward controller and feedback controller. The copy of the identifier is used as feed forward controller. The simulation result on two typical dynamic servo system indicates, the proposed controller can more improve tracking ability of servo system for dynamic signal. The time of the controller calculation can be reduced, thus, the requirement of real-time control is better satisfied.