Design of Stator Flux Observer Based on Neural Network for Induction Motor

The stator flux observer is a key part in the method of Direct Torque Control (DTC). However, the accuracy of the stator flux estimation directly affected the performance of DTC. The traditional induction motor stator flux observation method have been analyzed in This paper. And for the Shortcomings of existing methods, a on-line identification methods based on Radial Basis Function(RBF) have been proposed in the paper. First, the reference model of flux identification should be established according to induction motor u-n mathematical model under the static coordinate system. Then, a RBF neural network can be constructed on this basis. After self-organization learning, on-line identification of stator flux can be realized in the RBF neural network. System simulation has been carried out in Matlab / Simulink. The results show that: the identification method based on the RBF Neural network can improve the induction motor stator flux measurement accuracy, reduce the impact from the interference factors in observation process and the structure is very simple.