Wavelet based current signature analysis for motor health assessment

Abstract

Machines degrade as a result of wear and aging, which decreases their performance reliability and increases the potential for faults and failures. An approach is proposed for a health state detection and identification scheme for electric motors to achieve the robust performance prediction and failure prevention. The methodology uses discrete wavelet transform to analyze the detail local contents of the input signals through different levels of wavelet decompositions. The proposed approach uses motor stator current signature analysis as signal processing module in the algorithm. But the actual improvement in performance is achieved through the coefficients representation of discrete wavelet transformed signals. It is equipped with a multiple regression method to obtain the best and simplest correlation model between features and faults.