Residual Life Prediction Using Grey Correlation Analysis of Feature Selection

Abstract

Residual life prediction using monitoring data is an important method in reliability engineering. However, in the current performance degradation study, the selection of performance degradation characteristics is usually selected based on expert experience. In this study, a method using grey correlation analysis is applied to select the performance degradation characteristics as the health index of products. Further, the selected characteristics are analyzed with the linear Wiener process model. And the model parameters are estimated using the MCMC (Markov Chain Monte Carlo) method in the view of Bayes theory. Finally, a numerical example concerning the remaining life estimation of a certain satellite product is presented.