Fault Detection and Identification Using Deep Learning Algorithms in Induction Motors

Owing to the 4.0 industrial revolution condition monitoring maintenance is widely accepted as a useful approach avoiding plant disturbances and shutdown. Recently, Motor Current Signature Analysis (MCSA) reported technique in detection identification of individual multiple Induction (IM) faults. However, checking fault classification with deep learning models its comparison among themselves or conventional approaches rarely literature. Therefore, this work, we present induction motor faults MCSA three Deep Learning (DL) namely MLP, LSTM, 1D-CNN. Initially, have developed model Squirrel Cage MATLAB simulated it for single phasing stator winding (SWF) using Fast Fourier Transform (FFT), Short Time (STFT), Continuous Wavelet (CWT) detect identify healthy unhealthy conditions phase ground, Analysis. The impact on current presented time frequency domain (i.e., power spectrum). simulation results show that scalogram has shown good time-frequency analysis showing energy during conditions. This further investigated 1D-CNN) classification) improvement three-phase motor. By simulating various MATLAB, collected signature data domain, labeled them accordingly created 50 thousand samples dataset DL models. All are trained validated suitable number architecture layers. simulation, multiclass confusion matrix, precision, recall, F1-score obtained several result shows can be used Moreover, efficiently classify based time-domain signature. In models, LSTM better accuracy all other These employing motors very predictive avoid shutdown production cycle stoppage industry.