Comparison of different noise analysis methods for error detection on induction machines

Abstract

Increasing levels of noise and vibration are early indicators for an evoluting error or defect on an electrical machine. There are many different approaches in monitoring the health of an electrical machine, e.g. noise-, vibration- or current-based methods. This paper investigates and compares different methods for noise analysis. Each method provides an own representation of the analysed signal. Therefore, the presented methods are compared regarding their suitability in analysing electrical machines. In the following sections, first the theory behind all methods is presented. Later on, selected error states are re-enacted using an induction machine. Finally, utilizing measured noise and acoustic signals allows assessing the presented methods.