Reliability Monitoring for an Individual Product Based on Degradation Information Fusion Reliability Monitoring for an Individual Product Based on Degradation Information Fusion

Abstract

Degradation analysis has become very useful in reliability monitoring and maintenance schedule making for complex systems and highly reliable products with long lifetime. It is suggested, in practical engineering, that more attention should be paid to the time-varying health state and reliability index for key service-products from the viewpoint of safety and life-cycle cost. In this study, the degradation information of both product level and component level is combined and adopted to assessing the time-varying reliability of an individual product. A novel nonstationary independent increment degradation model including both deterministic and stochastic parameters is proposed to depict the statistical property evolution over time. The product-level degradation data and the component-level degradation data are incorporated for the maximum likelihood estimation (MLE) of model parameters. Then the reliability characteristics can be described by updating the stochastic model parameters based on product-level degradation information via a Bayesian method. Consequently, the remaining useful life (RUL) and the mean time to failure (MTTF) can be estimated and updated once a new online performance measurement is obtained. Therefore, the time-varying reliability can be monitored for an individual product