Advanced Prediction of Supply Faults in Three-Phase Induction Motors using Integrated Leakage Flux and Thermal Imaging with Machine Learning Methods

The predominant use of three-phase induction motors in high-power machine industries makes it an important facet of electrical systems. Over the years, it has been observed that the power supply of the motor also plays an important role in maintaining the health of the motor. Large or frequent fluctuations in the supply can increase the temperatures of the motor, leading to faults in the other parts of the motor and ultimately failing the machine. Leakage flux signature can identify different types of faults in induction motors due to vital changes at the fault frequencies. Thermal imaging can provide changes in the surface temperature of the motor due to incipient faults by indicating the hotspots through thermal signatures. Integrating the data from these two sensors, the modality of single-sensor unreliability reduces. Using machine learning models, Convolution Neural Networks (CNN), decision trees, backpropagation, and regularization along with dataset can increase the accuracy of fault classification. The methodology used in this paper can be utilized to predict the type of fault in any given three-phase induction motor with higher accuracy at incipient stages.