Sensorless Control of Brushless DC Motor Based on Wavelet Theory

Abstract

This article seeks to develop a sensorless drive technique for brushless machines based on wavelet theory that provides the advantages of separating low-frequency and commutation effects. The approach adopts two methods of position prediction. The first method employs self-inductance variation, as established with finite element analysis. The second is based on induced voltage and zero-crossing point estimation. The problem of starting is resolved by sensing inductance for the first method and by providing a look-up table for each direction of rotation for the second method. Both proportional-integral-derivative and fuzzy control algorithms are developed, and simulated current and speed-controlled performance predictions are obtained. Daubechies discrete wavelet analyses of experimental and simulated waveforms are obtained, with emphasis on commutation intervals. An algorithm is developed to predict commutation instants from wavelet results. The simulation model and its wavelet analysis match the experiments.