Induction motors fault detection using independent component analysis on phase current signals

Squirrel-cage induction motors are among most used rotary machinery in many industrial fields. Fault detection in early stages is in high relevance due to technical and economic issues, and broken bars are among the most common faults in induction motors. This paper presents an approach to carry out detection of this failure using as input the current signal measured from one of the three motors phases. Independent Component Analysis (ICA) is used over the Fourier-domain spectral signals obtained from the input and its autocorrelation function. A notable difference on the standard deviation over a region of interest in one output can be distinguished in the current signals obtained from damaged and healthy motors. Obtained results show a correct classification percentage of 95.3% in average.