Machine Learning-Based Classification of Stator Inter-Turn Faults Severities Using Time Domain Features and Principal Component Analysis

Stator-based faults are prevalent in Induction Motors (IMs) across various industries, including the electric utility sector. To enhance the safety and reliability of IMs, condition monitoring and fault classification play a crucial role. This paper proposes a data-driven approach for classifying Stator Inter Turn Faults (SITF) severities in IMs using 3-phase stator current signals. The application of Principal Component Analysis (PCA) on the time domain features is also demonstrated, yielding promising results that showcase the potential of a Narrow Neural Network (NN) for accurate SITF severity classification. Particularly, the narrow neural network classifier showed promising results with a classification accuracy of 98.23% on test-set experimental data. This study explores innovative fault classification methods, leading to advancements in component-level diagnostics and enhancing reliability. It lays the foundation for more efficient and reliable preventative maintenance strategies.