Condition Monitoring And Life Prediction Of Rolling Bearing Based On Cosine Similar Entropy

Rotating machinery is widely used in many industries,and it is particularly important to study its safety performance.Rolling bearings are the most widely used and most fault-prone components in rotating machinery,once the fault or failure occur to rolling bearings,which will cause the mechanical equipment work abnormally.The running condition and remaining life of rolling bearings are crucial for the safe and reliable operation of the mechanical equipment.The change of the running state of the rolling bearing needs to be tracked accurately and timely.This is helpful for judging the current operating state,predicting the development of the future operating state,evaluating the remaining service life of the rolling bearing,and also providing guidance for the maintenance of the rolling bearing and crane equipment.In this paper,the application of cosine similar entropy based complexity measurement for rolling bearing condition monitoring and remaining life were studied.The main work is as follows:(1)The degradation indicator of rolling bearing based on cosine similar entropy was developed.Firstly,cosine similar entropy algorithm was introduced and the influence of parameters was studied.Secondly,sample entropy and fuzzy entropy were compared by simulation signal analysis.Finally,cosine similarity entropy was applied to degradation characteristics representation of rolling bearing.The analysis result of experimental data indicated the effectiveness of cosine similar entropy and its superiority to sample entropy,fuzzy entropy and root mean square value.(2)The ensemble empirical mode decomposition was introduced and combined it with multivariate cosine similar entropy for rolling bearing condition monitoring and degradation trend prediction.Firstly,the multivariate cosine similar entropy and ensemble empirical mode decomposition algorithm were studied.Secondly,the ensemble empirical mode decomposition of the signal was carried out by simulation and experiment.The effective intrinsic mode function component was selected by the correlation coefficient criterion,and the multivariate cosine similar entropy value of the effective intrinsic mode function component was calculated as the rolling bearing degradation index.Finally,combined the extreme learning machine with index,the rolling bearing degradation trend was effectively predicted.(3)The cuckoo algorithm was introduced and applied to the parameter combination optimization of the variational mode decomposition method,and the residual life of the rolling bearing was predicted by combining the multivariate cosine similar entropy.The cuckoo optimization algorithm and the principle of variational mode decomposition algorithm were introduced.In the beginning,the cuckoo search-based variational mode decomposition was proposed.After that,by comparing the variational mode decomposition and the empirical mode decomposition,the superiority of the cuckoo search—based variational mode decompositionwas proved.So,this method was used to decompose the experimental data,and the optimal intrinsic mode function component was calculated by correlation coefficient calculation and the multivariate cosine similar entropy value was calculated.Finally,the extreme learning machine was used to predict the remaining life of the rolling bearing.In summary,the paper focused on the construction of cosine similar entropy in the rolling bearing degradation index,the effectiveness of state monitoring,and the application of related theory in degradation trend and remaining life.It provides the new technological way for the rolling bearings in the assessment of performance degradation and life prediction.