Bearing fault detection via Park's vector approach based on ANFIS

Abstract

In this paper, Park's vector transformation and frequency domain analysis for fault detection of induction motors are introduced. Then a smart approach based on Adaptive Nuero Fuzzy Inference System (ANFIS) that uses time domain features obtained from the Park's transformation of stator currents is proposed for fault detection. By the proposed method, a 1 mm hole on the inner race and two faults including 1 mm and 3 mm hole on the outer race, using experimental data is investigated. It is shown that using features derived from Park's vector modulus results in better performance compared to the features obtained from a single phase current.