Prediction of Stator Faults in Motor Using Discrete Wavelet Transform and Elastic Net Regression Through Sensor Signals

Thermal imbalance-caused stator failures in motors lead to curving or stretching of magnetic flux, which subjected magnetic stress to the stator laminated surface. Such stress can damage the stator, reduce efficiency, and lead to harmonics as it is directly proportional to variations in slot size. Magnetic stress is a major issue that is commonly overlooked by traditional motor defect diagnosis methods. This study introduces a method to predict Stator Faults (SF) during motor operation using Elastic Net Regression (ENR). SF is estimated based on mean and median values from multimodal sensor signals including temperature and Giant Magneto Resistance (GMR) processed through Discrete Wavelet Transform (DWT). Stator slot sizes are also analysed using microscope images. Experimental results reveal that slot size exceeding 20% of the standard size causes significant and stator damage, with the prediction method achieving approximately 93.9% accuracy.