Design of Neuro-fuzzy Controller Based on Dynamic Weights Updating

摘要

Neural and fuzzy methods have been applied effectively to control system theory and system identification. This work depicts a new technique to design a real time adaptive neural controller. The learning rate of the neural controller is adjusted by fuzzy inference system. The behavior of the control signal has been generalized as the performance of the learning rate to control a DC machine. A model of DC motor was considered as the system under control. Getting a fast dynamic response, less over shoot, and little oscillations are the function control low. Simulation results have been carried at different step change in reference value and load torque.