Research On Remaining Useful Life Prediction Method Of A Rolling Bearing Based On SVR

Rolling bearing is one of the important basic parts in rotating machinery,the safety accident can be triggered and then great economic losses and bad social influence can be caused.The remaining useful life(RUL)of a rolling bearing is the comprehensive reflection of its the damaged extent,therefore,it is significant for condition maintenarnce to predict the RUL of a rolling bearing accurately.Two models are established for the RUL predcition of a rolling bearing in the research.(1)Research on the feature extraction and feature reduction method of a rolling bearing.The method is proposed which combine envelope demodulation with mathematical morphology,the mathematical morphology feature is extracted.Meanwhile,the feature matrix is constructed based on the time domain,frequency domain feature index and the fractal dimension of mathematical morphology,and the feature reduction method is proposed based on multiple criterion effectiveness analysis-kernel principal component analysis(MCEA-KPCA).(2)Research on the prediction method of a rolling bearing running status reliability.In order to ensure the accuracy of the reliability prediction of a rolling bearing and increase the prediction step length,a rolling bearing reliability prediction method is proposed based on the fractal dimension of mathematical morphology and improved fruit fly optimization algorithm-support vector regression(IFOA-SVR).Firstly,the fractal dimension of mathematical morphology is regarded as the performance degradation state feature of a rolling bearing.Secondly,the prediction model is established using SVR and IFOA,and performance degradation feature is predicted,at the same time,the Weibull proportional hazard model(WPHM)can be established using the maximum likelihood estimation combined with IFOA,then the reliability model can be obtained.Finally,the prediction results are embedded in the reliability model,then the reliability of the bearing running state can be predicted,namely RUL.(3)Research on the prediction method of the ratio p of the bearing running time to the whole life time.Aim at the situation which the working condition and type of the bearing in training process are same as testing process except bearing individual,a RUL prediction method of a rolling bearing is proposed using MCEA-KPCA and combined SVR(CSVR).The weights of every criterion and SVR model can be determined adaptively,the bearing which is different from training process is tested,the p value of the bearing is predicted,the experimental results show that the RUL of the rolling bearing can be predicted accurately by the proposed method.