Abstract:

Residual lifetime distribution prediction of products is of great importance in maintenance, replacement and spare parts decision making. The majority of existing methods use only the degradation data of products themselves, so precision of the prediction results is hardly satisfying when the degradation data are few. This paper tackles the problem using a Bayesian method. The degradation data of the products the Wiener process with random effect are modelled. Using the Bayesian method, featuring online degradation data and historical lifetime data are fused to derive the posterior distribution and Bayesian estimates of degradation parameters. Residual lifetime distribution is deduced, thus improving predictive precision. An example, residual lifetime distribution prediction of metallized film pulse capacitors, is presented to show the validity of the proposed method.