Electric Vehicle Induction Motor Fault Classifications Using Thermal Image Temperature Matrix Index and Machine Learning

Induction motors (IMs), as electro-mechanical systems, find extensive utility across various industrial applications, serving as prime movers for systems of both small and large scales. Their widespread adoption owes to several merits such as exceptional reliability, high efficiency, and resilient performance. Now days its normally used in Electric Vehicles (EVs) due to low cost and less maintenance. To ensure secure and continuous operation IMs, current condition monitoring techniques rely on various sensors for diagnosis. Infrared imaging technology, a non-invasive method for industrial fault diagnosis, is commonly utilized to assess equipment status, particularly in harsh environmental conditions. In this study, Thermal Pixel counting (TCP) is employed to extract the feature vector from the thermal image of the motor. These data are utilized for training an initial Backpropagation Neural Network (BPNN), which is subsequently utilized to train Restricted Boltzmann Machines (RBM) for the classification of induction motor faults, such as airgap eccentricity, shaft bending, and cooling system failure.