Signal Analysis in Converter-Fed Drives Using Adaptive Neural Networks

Abstract

Frequency converter-fed induction motor drives were simulated using the EMTP-ATP (Electromagnetic Transients Program-Alternative Transients Program). The aim of the simulation investigations is to develop a feasible method for drive visualization and for diagnosis of different unsymmetric operations. Visualization of a converter-supplied drive by means of modal space phasors is a very useful and compact observation and diagnosis method. Fast and exact estimation of the frequency of distorted voltages and currents at the motor input enables us to obtain stationary space-phasors and to determine an appropriate sampling window for a DFT analysis. For this purpose, an adaptive neural network is proposed. Computer simulation results confirm the validity and performance of the proposed method.