基于马氏距离的竞争失效可靠性分析

A reliability analysis method based on mahalanobis distance (MD) was proposed for the reliability evaluation at competing risk involving both degradation-based failure and shock-based failure. Firstly, MD was used to transform the multivariate degradation data to one dimensional MD data, and the MD data was modeled by Wiener process with random effects. Then, two extreme shock processes were considered, the shock processes with Poisson process and the shock processes with a fixed time. The reliability models of the two situations were established respectively. Both the shift threshold and minimum magnitude of shock loads were considered synchronously in these dependent competing failure models. Finally, the effectiveness of proposed models was demonstrated by a fatigue crack increase example. Besides, sensitivity analysis was performed to assess the effects of model parameters on the reliability. Results show that, both the shocks with Poisson process and with a fixed time affect the analysis results of product reliability.