Detection of Inter-Turn Short Circuit in Stator Windings of Electric Machines Using Magnetic Symmetry Index and Machine Learning Methods
Vahid Rafiei, Mahshid Khoshlessan, Carlos Caicedo-Narvaez, Babak Fahimi
Abstract
Electric machines have been in the spotlight as the transportation industry, along with many other high-impact industries, undergo the electrification process. While researchers debate whether one geometry or configuration has a higher power density, higher efficiency, or easier manufacturability, it is agreeable that nearly all the competing electric machines are symmetric. Symmetry can be used to develop a health index, whereas asymmetric signatures point to anomalies that stem from manufacturing imperfections, problems related to rotor faults, or inter-turn short circuits in the stator winding. This study takes a Switched Reluctance Motor (SRM) and an Induction Motor (IM) and analyzes their polar magnetic symmetry to introduce a stator health index. Moreover, this symmetry index in conjunction with a discriminative classifier is used to determine and classify the fault severity. Finally, this study presents a viable method to determine the health (in terms of inter-turn stator short circuit faults) of the motor and shows that support vector machine (SVM) and XGBoost can successfully classify the severity of an inter-turn short circuit fault for IM and SRM, respectively.