Abstract: A new control structure for load simulator to improve its performance is presented, in which there are three RBF neural networks applied, one is taken as NN PID controller, the second is taken as compensator for adjusting anti-disturbance coefficient dynamically, the third one is use for identifying the plant (NNI). The neural network PID controller is designed by requirement for steady, which overcome the shortcoming of the only neural network controller. A RBF neural network is used to identify the plant, in which the Jacobian matrix is got to use for adjusting disturbance-reduced parameter in the compensating neural network. The simulation and experiment are carried out, which show a good result. This method is of validity and robustness.

Key words: Compensation of extraneous force, hydraulic load simulator, Neural network controller Intelligent PID