Quasi 1D CNN-based Fault Diagnosis of Induction Motor Drives

Application of machine learning in the area of fault diagnosis of induction motor is somewhat popular. However the major challenges in this domain are restricted by the handmade statistical features, which limit the performances of the classifiers immensely. On the other hand deep learning, a new era of machine learning, has opened a new horizon, where self-synthesized features from the raw signals have already proven their superiority in various real life applications. This paper has exploited this merit of deep learning and reports a preliminary findings in motor fault detection using novel Quasi 1D Convolution Neural Networks (Q-1DCNN). The computation is fairly faster as compared to 2D CNN and the generalization of the results are encouraging.