A radial basis function neural network based efficiency optimization controller for induction motor with vector control

In this paper, a efficiency optimization controller is studied for a vector controlled induction motor. The optimum flux-producing current is obtained utilizing radial basis function neural network (RBFNN). Comparing with the conventional neural network, the radial basis function neural network possesses characteristic of simple structure, fast convergence and strong generalization, it is suitable for real-time control. Considering the change of iron core loss resistance due to flux and frequency, a precise dynamic model of induction motor is built, and RBFNN is trained based on this model. When induction motor is in steady state, the trained RBFNN is used to optimize flux-producing current; while when induction motor is in the transient state, the flux-producing current recovers the rated value for ensuring the fast response. The simulation is done with Matlab/Simulink and the proposed method of control is realized adopting TMS320LF2407A. Simulation and experiment results show that the efficiency optimization is significant and speed response is rapid.