An effective approach for electric motor fault diagnosis using deep learning

Induction motors have versatile applications across various industries. However, during their integration into different systems, they can be susceptible to a range of failures such as broken bars and interturn faults. To mitigate the risks of unforeseen motor breakdowns, this study introduced an Artificial Neural Network (ANN) based fault detector to assess the severity of fault conditions. The primary goal is to enhance the reliability and longevity of induction motors by promptly identifying potential issues. In this proposed model, Levenberg-Marquardt back -propagation algorithm is utilised for training and the ANN was subjected to testing under both healthy and five distinct fault conditions of the electrical machine.The results obtained from the experimentation phase are promising, revealing that the proposed ANN topology exhibits a noteworthy accuracy level of around 96%. This accuracy surpasses that of the pre-existing topology, indicating a significant advancement in fault detection capability.