Inner race bearing fault detection using Singular Spectrum Analysis

Abstract

A novel method to diagnose the bearing fault is presented. The proposed method is based on the analysis of the bearing vibration signals using Singular Spectrum Analysis (SSA). SSA is a non-parametric technique of time series analysis that decomposes the acquired bearing vibration signals into an additive set of time series to extract information correlated with the condition of the bearing. Information in terms of time-domain features extracted from the SSA processed signal has been presented to a neural network for determination of inner race bearing fault. The result shows the effectiveness of the proposed method.