Research on a rolling bearing health monitoring algorithm oriented to industrial big data

In view of big data of rolling bearing, traditional feature extraction methods will cause some useful information loss and the existing bearing performance degradation index hardly presents its actual operation state. A method (RMT-PCA) based on random matrix theory (RMT) and principal component analysis (PCA) is proposed to overcome these shortcomings, which utilizes rolling bearing health monitoring data to construct a random matrix model; 14 feature indexes are extracted and constructed by using random matrix theory; the PCA algorithm extracts useful information from multiple feature data and finally constructs a fusion feature index to assess bearing's degradation. An application research is carried out by using bearing datasets of American IMS and IEEE PHM 2012 Prognostic challenge, and compared with other intelligent algorithms, the results reveal that the RMT-PCA algorithm is more sensitive to the early anomaly and can accurately and truly reflect the actual degradation process of the bearing.

Automated summary

The RMT-PCA method utilizes rolling bearing health monitoring data to build a random matrix model, extracting 14 feature indices and constructing a fusion feature index through PCA algorithm to assess bearing degradation. Experimental results demonstrate that the RMT-PCA algorithm is more sensitive to early anomalies and accurately reflects the actual degradation process of the bearing.