Remaining Life Predictions of Bearing Based on Relative Features and Support Vector Machine

Abstract

A new prediction method is proposed based on relative features and support vector machine to estimate the bearing remaining life under limited data conditions. To eliminate the redundancy and relevance within features, principal component analysis (PCA) was applied to obtain the relative features, which could reflect the running states and degradation trends of bearings. Then, the relative features are input into the support vector machine. The bearing residual life prediction model is constructed based on the relative features and support vector machine. The field measured signals are used to verify the effective of the proposed method. The results show that the proposed prediction method can obtain accurate prediction results under small sample conditions.