Various Faults Classification of Industrial Application of Induction Motors Using Supervised Machine Learning: A Comprehensive Review

Induction motors are important to the industrial sector, acting as the backbone to various processes and machinery in several fields. In fact, with the huge consumption of electrical energy by the industry, approximately 90% of industrial processes rely on these motors. However, these devices are susceptible to various faults such as electrical and mechanical faults, which can negatively impact their operation and significantly affect motor performance. The main occurrences of these faults are outer and inner race faults in the bearing, eccentricity misalignment due to broken rotor bars, and short-circuit severity issues in the stator winding. Recently, the growth of artificial intelligence has infiltrated a variety of fields, including fault detection in induction motors. In current literature, there are a number of papers that address all these faults using different methods, and this paper compiles the information from the written works for ease of access. Machine learning algorithms are a set of data-driven rules that are able to classify specific faults in induction motors, which will be explained further in this review paper. This paper presents a comprehensive review of recent techniques proposed in the literature for bearing, stator winding and broken rotor bar faults with machine learning algorithms focused on fault detection for numerous faults. The review thoroughly examines the advantages and disadvantages of these online methods and provides a detailed comparison across various aspects. Finally, the study identifies the major challenges and research gaps in these techniques.