RVM-based adaboost scheme for stator interturn faults of the induction motor

Abstract

This paper presents an AdaBoost method based on RVM (Relevance Vector Machine) to detect and locate an interturn short circuit fault in the stator windings of IM (Induction Machine). This method is achieved through constructing an Adaboost combined with a weak RVM multiclassifier based on a binary tree, and the fault features are extracted from the three phase shifts between the line current and the phase voltage of IM by establishing a global stator faulty model. The simulation results show that, compared with other competitors, the proposed method has a higher precision and a stronger generalization capability, and it can accurately detect and locate an interturn short circuit fault, thus demonstrating the effectiveness of the proposed method.