Inference on Accelerated Life Testing for One-Shot Device with Competing Risks

This article deals with modelling and analysis of the competing risks for a one-shot device under a constant stress accelerated life test. In a reliability analysis of a device, it is important to be able to identify the main causes of failure. Therefore, a competing risk model is generally used. We consider this model in two modes: observed and masked causes of failure. The data obtained from one-shot device testing are missing in fact. For this reason, the EM algorithm along with the Fisher scoring method are used to estimate the model parameters. An accelerated life test is also used to shorten the time and cost. In addition, in order to accurately estimate the product reliability, the test design is finally optimized. Based on the simulated study, it is concluded that the EM algorithm and the bootstrap confidence interval are more accurate than the other methods. Also, shortening the test length leads to achieve an optimal test design.