Neural network control of electric machines for transportation systems

Ac electric motor drives are widely used in applications of electric vehicle and subway transportation. The dynamic performance of ac motor control strongly depends on model parameter accuracy. As a result traditional control scheme can´t achieve good performance under uncertainty parameters. In this paper a compound gradient vector based neural network algorithm is investigated and applied in induction motor drive control. The convergent analysis of the algorithm indicates that because the compound gradient vector is employed during the weight update, the convergent speed of the algorithm can outperform that of the BP algorithm. Some simulations have been carried out and the results verify that satisfactory convergent performance and strong robustness are obtained in ac motor drive control involving uncertainty parameters.