Time domain diagnosis of multiple faults in three phase induction motors using inteligent approaches

The three-phase induction motor is one of the most employed equipment in industrial premisses. Despite of its reliability and robustness, these machines can present faults due to the operation time, harsh operating conditions, voltage unbalance, among other factors. In this work, a methodology for intelligent diagnose of multiple faults in induction motors by using a discretization of currents and voltages amplitudes signals in the time domain is proposed. Three types of intelligent classifiers are employed to proper diagnose motor faults: artificial neural network type multilayer perceptron (ANN/MLP), algorithm k-nearest neighbour (k-NN) and support vector machine with sequential minimal optimization (SVM/SMO). The investigated faults are related to stator short-circuit, broken rotor bars and bearing defects. Experimental results are obtained with data gathered from a 1 hp motor under varied load and unbalanced voltage conditions. The MLP and k-NN classifiers are highlighted with accuracy above 89%.