Induction Motor Identification from Acoustic Noise Spectrum by a Covariance Subspace Algorithm

Abstract

In this paper, we study identification of induction motors by using a recently developed covariance-based subspace algorithm from sound measurements. The sound data are collected by an array of five-microphones placed hemispherically around motors in a reverberant and noisy room. It is demonstrated that mechanical faults, frequently encountered in induction motors, are isolated by the proposed identification algorithm. The estimated acoustic noise spectra can be utilized in the development of preventive maintenance programs for induction motors in service