Condition monitoring of induction motors using combined PSD based wavelet decomposition and selective weighting using spider web plots

Abstract

Condition monitoring of machines has its roots in the human ECG analysis for detecting cardiac arrhythmias. Condition monitoring in industry is desirable for increasing machinery availability, reducing consequential damage, and improving the operational efficiency. This is very significant in industries that use heavy duty machines for various processes. E.g.: A monstrous three-phase AC induction motor to drive a city water supply pump or very big, high power motors used in mills, huge generators for generating power in hydel plants etc depict a few of them. Also, for safety and economic considerations, there is a need to monitor the behavior of motors working in critical production processes as well. This paper demonstrates how the condition of an induction motor can be monitored by the analysis of the acoustic signal that represents the non-stationary vibration data. The analysis have been done using various signal processing algorithms and a new and better classification technique has been developed using the combined wavelet based PSD analysis and Spider Web Plots. The proposed method appears to give better performance.