An adaptive controller based on neural networks for Motor-drive load simulator

A based Radial-Basis-Function (RBF) networks adaptive control scheme is proposed for Motor-driven load simulator in the paper. The system is highly nonlinear and includes delays in the control loop and extraneous force disturbance from load. These, factors have many bad effects on the tracing accuracy and dynamic performance of the simulator. Contrasted with the conventional-BP-networks-based controllers, the proposed algorithm is more efficient without the local-minimal problem. This controller is robust, efficient and simple. The structure invariance principle is used to reduce the extraneous force, but in fact, this method does not work well due to the modeling error and the speed error produced by the speed sensor. The RBF neural networks can be used in adaptive controller for simulator, and the outputs of the neural networks are used as the parameters of the controllers to compensate for the effects of mentioned factor. In this paper, the key problem discussed is to design a load simulator with high tracing accuracy and wide bandwidth. The simulation results presented in the paper shows that the designed controller provides good control performance and adaptive compensation of the extraneous force.