Research on robustness of BP neural network based inverse model for induction Motor Drives

Abstract

Since the well design of inverse model is very important for inverse decoupling control of induction machine (IM) drives, the robustness of BP neutral network based inverse model is deeply researched by simulation experiments in the paper. First, the dynamic model for IM drives is constructed with the use of state space theory. Second, the inverse model for IM drives is set up by inverse system theory. However, the analytic inverse model is hardly applied in the engineering field since it excessively depends on the parameters. Third, an artificial neural network (ANN) based inverse model, which synthesizes artificial intelligent method and analytic method, is suggested in this paper. To accelerate the convergence speed of ANN and enhance its generalization ability, the nonlinear parts are realized by the analytic expressions and the corresponding results act as the inputs of ANN so that the complex-nonlinear mapping relation become a simple-linear mapping and the structure of ANN is simplified by the greatest extent. A three-layered feed-forward ANN with 12-10-2 structure is introduced to approach the inverse mode of IM drives. Lastly, the robustness of ANN based inverse model is verified by comparative experiments in the case of parameter variance.