Just-in-time Detection of Stator Short Circuit Faults in Electric Machines

Electric Machines are often considered the workhorse of today’s industry. The global electric motor sales market size was valued at USD 126.9 billion in 2018 for applications ranging from aerospace and electric vehicles to air conditioning equipment and household appliances. Such application diversity and market size illustrate the importance of continuous operation of electric machines with minimized downtime due to faults. As electric machines fail, down-time can directly translate to economical loss in consumer applications and critical loss in sensitive applications such as medical and aerospace products. Stator winding failure accounts for approximately 38% of all motor faults. Aging due to operation, unbalanced input/load, and partial electric discharge are some of the causes that lead to insulation damage in the stator windings, eventually creating an inter-turn short-circuit fault. When two neighboring conductors come into contact, electrical, magnetic, and therefore thermal characteristics of the stator change in an asymmetric manner. This dissertation studies asymmetric thermal and magnetic signatures and develops a stator health index that successfully identifies and classifies stator anomalies in the form of short circuits and manufacturing imperfections. Thermal and magnetic data is acquired by taking a Big-Data approach, where the data is aggregated and uploaded to the cloud for storage, analytics, and machine learning modeling. The detailed system for detecting stator anomalies is presented and validated.