Research on the Identification of Stator Resistance Based on the Theory of the Wavelet Fuzzy Neural Network

Based on temperature of stator winding and its rate of variety have characteristics of nonlinear and time frequency, which is hard to build precise mathematical model. A novel approach to identifying the induction motor stator resistance on line is presented. This method is based on the theory of the wavelet transform fuzzy neural network identification algorithm. Using the time frequency location characteristic and multi-scale ability of wavelet transform, and the recursive orthogonal least squares algorithm, this paper proposes the modified "Givens" rotations which avoids orthogonal decomposition of complex matrices. The result of simulation experiment show that using this method can indicate the result of stator resistance and its time frequency, and this effectively improve the low-speed performance of the direct torque control system.