Double Layer Feature Selection and Feature Fusion Method for Rolling Bearing Based on AFF-AAKR

As the basic part of mechanical equipment, rolling bearing plays an important role in rotating machinery. Therefore, condition monitoring and remaining useful life (RUL) prediction of rolling bearings are particularly important in industrial production. Accurate feature extraction and efficient feature fusion are the two key steps of traditional data-driven RUL prediction method. This paper proposes a double layer feature selection method. Primary screening of features is to select the sensitive subset which is sensitive to the degradation trend through the comprehensive index sorting. Re-selection of features based on hierarchical clustering and mutual information can further eliminate redundant features from sensitive feature subset, and finally get the optimal subset. Then a feature-based fusion method based on adaptive feature fusion method (AFF) and auto association kernel regression model (AAKR) is proposed to construct health indicator which can reflect the degradation trend of rolling bears.