Design criteria for planning multiple stresses accelerated life test

In general, there are three types of methods to arrange the stresses combinations in optimal design of multiple CSALT: Escobar-Meeker method, orthogonal design method and uniform design method. In this paper, these methods are compared through computer experiments, and the design criterion for multiple stress accelerated life test is proposed. Considering the linear-extreme value statistical model, the plans are obtained based on the above three methods under V-optimality. These plans are compared from the following three indicators: the estimation precise of the p-th quantile of life distribution under normal stress, the robustness of optimal test plans to misspecified model parameters and the estimation precise of model parameters. The results show that, the plans obtained from Escobar-Meeker method is the best, and followed by orthogonal design and uniform design; Escobar-Meeker plan with five test points is the best plan compromising the three aspects. The research in this paper provides theoretical support to select the best methods to arrange the stresses combinations in multiple stress accelerated life test planning.