Reliability analysis of degrading systems based on time-varying copula

Abstract

Modern engineering products usually have complex structures and different functions, which means that products may have multiple degradation failure mechanisms. At present, the study of multivariate degradation reliability is mostly limited to copula function with constant parameters. For high-reliability products with multiple degradation processes, a bivariate accelerated degradation reliability estimation method based on time-varying copula is proposed. For accelerated degradation data, the marginal degradation model is constructed by using the nonlinear Wiener process with random effect, and the influence of environmental covariates is considered in combination with the acceleration model. The joint-dependence structure between two degradation processes is described by time-varying copula. A Simulation study validates this approach and the consequences of model misspecification are evaluated too. Furthermore, a real-world example is presented to demonstrate the effectiveness of the method. The results show that this method has some advantages over the existing methods.