Research on the Fuzzy Neural Network PID Control of Load Simulator Based on Friction Torque Compensation

To decrease the influence of friction on torque tracking accuracy and improve the rapidity of system response when load simulator works at low frequency and low speed, a novel method based on fuzzy neural network (FNN) PID controller and friction torque compensation is put forward. The FNN PID consists of FNN and neural network (NN) PID. The parameters of the controller were optimized by the mixed learning method integrating of offline genetic algorithm (GA) and online error back propagation (BP) algorithm. The friction torque model is identified by LuGre model. The loading motor is a double-stator motor in which the outer stator system serves as compensating the friction torque and the inner stator system as loading torque. Simulation results show that the control system has good dynamic and static performance.