Application of NN-PI Controller in Direct Current Motor Servo System

In motion control, static friction coming into being nonlinearity in actuator results in the tracking error when speed or moment turns around. It is difficult to eliminate as usual. Neural network can self-learn and approach to any nonlinear function at any given precision, so it can be used to deal with such problem. In this paper the friction phenomenon is analyzed and the friction model given. A friction compensation method based on BP neural network is presented. Neural network identifier and neural network compensator are trained with input and output data acquired from servo motor and compensator is connected with PI adjustor in parallel. So measurement results are compared with results adopting fuzzy-PI algorithm. Application in DC motor servo system indicates that this method improves further the system responding speed, and improves the tracking precision accordingly.