ABSTRACT
Monitoring of mechanical condition of electro-mechanical circuit breakers as reported in [1], [2] and [3] reflects the necessity of a noninvasive method for predictive maintenance. By far the most common source of malfunction of circuit breakers is due to mechanical faults that are dependant on the number of operations.
In attempting to provide an alternative method for predicting the mechanical condition we have postulated that instead of using spectral information [1] we would simply make use of the original time domain signal. For the pattern recognition process a backpropagation trained multilayer perceptron was implemented.
From results obtained it appears that an accurate classification of the vibration signature of an impulsively loaded mechanical component can be achieved. Not only can faults be detected but a reliable indication of the specific type of abnormality can also be achieved. This type of condition classifier will be very effective in early fault detection and prediction of mechanical failure in large electro-mechanical circuit breakers.
