The Optimal Designs And Statistical Analysis Under Accelerated Life Tests Under Multivariate Exponential Distribution With Dependent Competing Risks

The competing risk model is commonly used in the reliability analysis. Among those papers discussing statistical analysis of competing failure data, most of them assume independence for failure mechanisms. To reflect and assess the true depen-dence, we use M-dim Gumbel Multivariate Exponential Distribution to build model when the M failure mechanisms have correlation and research its statistical model. First, under the constant-stress and step-stress, the basic process of the accelerated life tests of the multivariate exponential distribution is introduced and correspond-ing parameters are estimated. Next, given the accelerated equation of k stress withκunknown parameters, the optimal designs of the accelerated life tests of typeⅠand typeⅡunder the constant stress and step stress are respectively studied according to D-Optimal and V-Optimal.