Analytical investigation of torque and flux ripple in induction motor control scheme using wavelet network

摘要

An effective scheme of parameter identification based on wavelet neural network is presented for improving dynamic performance of direct torque control system. The wavelet transform is localized in time-frequency domains, yielding wavelet coefficients at different scales. This gives the wavelet transform much greater compact support for analysis of signals with localized transient components. The input nodes of wavelet neural network are current error and change in the current error and the output node is the stator resistance error. To fulfill the network structure parameter, the improved least squares algorithm is used for initialization. The stator flux vector and electromagnetic torque are acquired accurately by the parameter estimator once the instants are detected. This function can make induction motor operate well in low region and can optimize the inverter control strategy. The simulation results show that the proposed method can efficiently reduce the torque ripple and current ripple.