Temporal Novel Approach for Bearings Faults Detection and Isolation in Wind Energy Conversion Systems

Abstract

This paper presents bearings faults detection and isolation system for a wind energy conversion system (WECS). For this, and contrary to the traditional methods often used and based on the frequency and/or the vibration analysis of generator signals, this novel approach is based on the temporal analysis of electrical signals (current or voltage) of the generator. The method is based on the observer scheme, composed of a time-varying Kalman filter and the strategy of the mean-residual to generate new residual capable to detect and quantify all bearings faults types. The proposed system has been validated on signals of a doubly-fed induction generator and the simulation results approve its efficiency.