MDCCS Based Multistage Life Prediction of Slewing Bearing with a Novel Performance Description: an Improved Variational Mode Decomposition Approach

Residual Useful Life (RUL) prediction plays a vital role in the domain of rotating machine condition monitoring. There have been extensive researches on the small bearings’ RUL prediction. Meanwhile, it is relatively less for slewing bearings which are used as connections between slewing systems and endure heavy load and harsh working conditions, its reliability of operation guarantees the safety of whole machines. Traditional condition monitoring of rotating machine mainly relies on vibration signal, slewing bearings’ own unique characteristics such as: large size, low speed, heavy load and complicated degradation process make life prediction very difficult and quite different from small bearings. Although many scholars have devoted their efforts to developing suitable slew bearings’ life prediction models, few have taken their intrinsic characteristics into consideration. To address this issue and improve the precision of RUL prediction, this paper first proposes High Dimensional Mode Signal Group (HDMSG) which is generated by Multi-Source Optimized Variational Mode Decomposition (MSOVMD). It is designed to contain Intrinsic Mode Functions (IMF) from diverse vibration signals to make a more comprehensive description of slewing bearings’ degradation and enhance damage signal. Degradation features extracted from HDMSG, temperature and torque signal are selected and fused afterwards. Secondly, considering the complex damage or degradation of slewing bearings, Maximum Differential Coefficient-Clustering Segmentation (MDCCS) is proposed and introduced as Health Stage (HS) division method which takes both advantages of multi-physical signals’ trend and clustering algorithm. It is worth noting that two accelerated run-to-failed experiments under completely different loading process were conducted for verification. Comparisons demonstrate the superiority of the alternatives of vibration signal: HDMSG and MDCCS based HS division methods.