Prediction Of Performance Degradation Trend Of Rolling Bearing Based On Hypersphere Support Vector Machine

Rolling bearing,as an indispensable part of rotating machinery,its performance degradation directly influence the operation state of the whole equipment.Through the study about the prediction of performance degradation trend of rolling bearing,the degradation state is received timely and accurately,which provides supports for the subsequent reliability assessment and life prediction,so as to avoid the overall damage of equipment in advance due to the damage of bearing,and the large economic losses and catastrophic accidents.Therefore,the deep study about the prediction of performance degradation trend of rolling bearings is quite necessary.In this paper,the rolling bearing is taken as the study object to make research on the preprocessing of vibration signal,the establishment of performance degradation indexes,and the model of prediction of performance degradation trend.Against the problem that a large amount of noise is mixed into the vibration signal of rolling bearing for the increasing errors of subsequent prediction,the combined signal preprocessing method based on EEMD-improved generalized morphology is put forward in this paper,in which the noise is effectively filtered out,and the integrity of original signal is greatly kept at the same time.After the compared experiment of two single signal processing methods,the effectiveness of the above method is proved with the noise reduction results.A construction method of multi-feature performance degradation indexes based on multi-core optimization KPCA is also put forward in this paper,which further solves the problem that the single performance degradation index can not fully reflect the performance degradation trend of rolling bearing,the redundant original multi-dimensional feature index information and so on.Moreover,after the related testing of the lifetime data,the performance degradation trend indexes with high sensitivity and overall degenerative reflecting are received.The trend prediction model based on hypersphere supported vector machine is put forward in this paper,and the multi-core optimization on its kernel function is made,and the parameters are optimized with the adoption of improved gray-wolf algorithm.The improved gray-wolf optimization algorithm-multi-core hypersphere supported vector machine models is constructed to further solve the problems of poor performance degradation recognition effect due to the isomerism and uneven distribution of vibration data.The accurate performance degradation trend prediction of rolling bearings is conducted through testing the lifetime data of it.In the end,the performance degradation trend prediction system of rolling bearing is developed in this paper through adopting the VC ++ software development platform based on the above study results,so that the effectiveness of this method from the angle of practical application is proved.Through the study about the prediction of performance degradation trend of rolling bearing,its performance degradation state can be predicted in advance to avoid the large economic losses and catastrophic accidents due to the damage of bearing,and provide important reference for the subsequent predictive maintenance and lifetime prediction.