Detection of faults in induction motors using artificial neural networks

When faults begin to develop, the dynamic processes in an induction motor change and this is reflected in the shape of the vibration spectrum. Thus, one can detect and identify machine faults by analyzing the vibration spectrum by examining whether any characteristics frequencies appear on the spectra. Here, the authors describe how fault detection and identification using such a vibration method on a induction motor was accomplished using a simple neural network program. Two machine faults, of bearing wear and unbalanced supply fault, are simulated and tested. Acceptable results are obtained and faults are classified accordingly.