Analysis for partially accelerated dependent competing risks model with masked data based on copula function

In the reliability analysis of competing risks model, it is usually assumed that the components are independent and failure causes are observable. However, this assumption may be unrealistic in some applications. In this article, based on the tampered failure rate (TFR) model, we introduce a constant-stress partially accelerated life test (PALT) model of dependent competing risks with masked data. The dependent structure is described by Gumbel copula function and the life distribution of components is described by the Weibull distribution. The maximum likelihood method is used to estimate the dependent parameter and unknown parameters of the model, so as to derive the estimation of the reliability function (RF). Confidence intervals (CIs) of the model parameters are also obtained via asymptotic normality theory and bias-corrected percentile bootstrap (Boot-BCP) technique. Simulation results show that the dependent model with masked data improves the estimation accuracy of system reliability compared with the independent model. Finally, a real data analysis indicates that the statistical inference model and methods have good performance.