Remaining Life Prediction Of Rolling Bearings Based On Weibull Proportional Risk Model

Bearing is one of the important parts in mechanical equipments,and the stability of its running state is directly related to the reliability of mechanical equipments.Based on the historical data of the entire life cycle of the rolling bearing,the performance degradation index of the rolling bearing is extracted to establish the remaining life prediction model,the failure rate and reliability of the model are analyzed,the failure threshold is determined,the prediction of the remaining life of the rolling bearing is implemented,and the reduction of the accident rate The health management of equipments is of great significance.Aiming at the accuracy of the remaining life prediction of rolling bearings,the research on feature extraction,feature dimension reduction and remaining life prediction based on vibration signals is carried out step by step.The main research contents are as follows:Firstly,aiming at the problem of bearing residual life prediction under a small sample,the multi-domain characteristic information that can reflect the degradation state of rolling bearings is extracted.For the traditional remaining life prediction,a single time-domain or frequency-domain indicator is used for analysis and prediction.The local and global characteristics of the vibration signal cannot be taken into account.The data-driven empirical mode decomposition is used to extract the time-frequency domain features.And frequency domain feature parameters,Fisher scores are used to remove insensitive features,and a multi-domain feature parameter set that better reflects bearing performance degradation is selected.Secondly,a Laplacian feature dimension reduction method based on fuzzy C-means is proposed.As traditional linear dimensionality reduction methods cannot reveal the internal non-linear structure of the data and kernel principal component analysis based on kernel functions,it is easy to cause different types of sample points to overlap each other in the feature space,and it is impossible to achieve the nonlinear dimensionality reduction of high-dimensional data.A manifold reduction method based on Laplacian feature maps is used to reduce the feature dimensions of the target data using fuzzy C-means to classify the degradation stages.The Laplacian dimension reduction method based on fuzzy C-means is verified through experiments.Then,the remaining life prediction model of rolling bearings based on Weibull proportional hazard model is established.Aiming at the shortcomings of the trend prediction method of the Weibull proportional hazard model,a trend prediction method based on a differential autoregressive moving average model is proposed,and the failure rate and covariates are trend predicted,the failure threshold is set,and the method is verified by experiments.Finally,the full life test data of the rolling bearing is used for test verification.The comparison of test results shows that the method proposed in this paper can accurately predict the failure rate and remaining life of rolling bearings.