研究期間：10108~10207;In this proposal, we will discuss the reliability analysis of a series system under accelerated life tests. In a series system, the system fails if any of the components fails. It is often to include masked data in which the component that causes failure of the system is not observed. In most existing work the life time distributions of the components are assumed to be independent. However since different components are connected within the same system, the system operation may cause correlation among the components so that their lifetimes are indeed dependent. On the other hand, when the proportion of the masking data is high, it makes the maximum likelihood approach fail due to lack of information, especially in estimating the correlations. In addition to the EM algorithm, data imputation might also be needed in order to capture the real data structure. Bayesian approach is an appropriate alternative in such case. In this proposal, we will study both the maximum likelihood and the Bayesian approaches under the above framework. On the other hand, statistical inference of the commonly used lifetime distributions such as the exponential distribution, Weibull distribution, and lognormal distributions are well developed in univariate case. However, we will focus on multivariate lifetime distributions. How to define and/or simulate the data from other multivariate distributions is another issue to be considered in this proposal. We will also plan to propose some further statistical inference on the correlation among the components.