A novel neuro-classifier using Multiscale Permutation Entropy for motor fault diagnosis

Accurate and reliable fault detection in three phase induction motors is of great importance from economical perspective. This paper deals with the modeling of five different stator faults, viz. Single Phasing, Single line to ground fault, over-voltage, under-voltage and voltage unbalancing. As part of data acquisition, stator phase current values are recorded during healthy condition as well as during various faults. Multiscale Permutation Entropy is introduced to extract the statistical data from the phase current signal. The extracted information is used to train a Time-Delay Neural Network which acts as a fault classifier. The accuracy of prediction and fault classification is ascertained in terms of two statistical parameters namely, Mean Absolute Percentage Error and Root Mean Squared Error. The proposed synergy of Multiscale Permutation Entropy and Time-Delay Neural Network proves to be a highly effective fault diagnosis platform for on-line implementation.