Data analysis of progressive‐stress accelerated life tests with group effects
Liangliang Zhuang, Ancha Xu, Binbing Wang, Yuguo Xue, Songzi Zhang
Abstract
Progressive-stress accelerated life testing (PSALT) is a special type of experiment that tests the lifetime of a product with continuously varying stress levels. Due to the limitations of testing equipments and costs, the lifetime data collected by PSALT are usually censored and have group effects. In order to deal with the two characteristics in the data, this paper presents a novel PSALT model with group effects under progressive censoring. Two-stage and Gauss-Hermite quadrature methods are proposed to estimate the model parameters, while the interval estimates are constructed by bootstrap and the asymptotic theorem, respectively. Simulation studies are conducted to compare the proposed model with the traditional models without group effects in terms of the relative bias and root mean squared error under different scenarios. The results show that the proposed model can detect group-to-group variation, and that the models without group effects will result in large biases for estimating the characteristic lifetime of the product. Finally, the proposed model is validated by a real dataset.