A New Strategy for Induction Motor Fault Detection Based on Wavelet Transform and Probabilistic Neural Network

Abstract

Broken rotor bar faults cause of motor malfunction and reduction of the life cycle. For the safe and appropriate performance of induction motors, the motor fault detection is a critical issue. This paper presents a new strategy for the broken rotor bar fault detection of the induction motors. Finite element method (FEM) is used for accurate fault modelling and the flux density under broken rotor bar faults has been comprehensively analyzed. Moreover, a new rotor bar fault detection method based on probabilistic neural network (PNN) and wavelet transform is presented. The proposed approach uses the stator current signal amplitude samples in the time-frequency domain to extract the appropriate coefficients where they are considered as inputs to a PNN. The output of the PNN classifies the status of the rotor to a healthy or faulty condition. The performance of the proposed method is verified using numerical simulation.