Fault Detection and Diagnosis of Faulty Power Connections of Induction Motors Based on SVM and KNN Model

Induction motors (IM), particularly squirrel cage types, are integral to various industrial applications because to their simplicity, durability, and low maintenance needs. These encompass electric vehicles, machining apparatus, and railway propulsion systems. Despite their numerous advantages, stator windings and bearings are the most common failure points in induction motors due to the harsh operating conditions. Defective power connections in induction motors are a significant issue as they severely impair performance. The diagnostic system developed to address this issue comprised feature extraction, model training, and preprocessing. Data quality was ensured during the preprocessing phase through the implementation of various procedures. To prevent linear mapping of input data into a less distinguishable feature space, the kernel method was employed with ICA during the feature extraction phase. The SVM-KN model, which integrates two prominent machine learning methods, was employed for classification purposes. In experimental evaluations, the SVM-KN model surpassed solo SVM and KNN models, achieving a classification accuracy of 95.74% in fault power situations. The findings indicate that the amalgamation of SVM and KNN significantly enhances the diagnostic proficiency for defect detection. The proposed technology demonstrates potential for enhancing the operational safety and dependability of induction motors in industrial applications.