Abstract: A degradation condition monitoring method based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and principal component analysis (PCA) for rolling bearings is proposed, which considers the problems of weak fault signals and diverse fault modes. In this paper, intrinsic energy features and failure frequency features are respectively extracted from the preprocessed original vibration signals of rolling bearings using CEEMDAN algorithm and Hilbert-Huang transform. Then, a health index of the rolling bearings is obtained through the feature fusion using spearman rank correlation coefficient and PCA. Finally, the degradation condition of the rolling bearings is identified by the analysis of the monotonicity, robustness and correlation of the health index. The conclusion is drawn by a case study that the proposed method can be applied to accurately and timely identifying the degradation condition of the rolling bearings.

Key words: rolling bearing, empirical mode decomposition, principal component analysis (PCA), condition monitoring