中文摘要：

本文研究了定时和定数截尾情形CE模型下Weibull分布场合步进应力加速寿命试验的Bayes估计.利用加速系数和加速方程将各种加速应力水平下的尺度参数换算为正常应力水平下的尺度参数,从而获得含正常应力下尺度参数的似然函数.在参数先验的选取时,尺度参数和加速系数分别取共轭先验和无信息先验,当形状参数m〈1和m〉1时分别取Beta分布和Gamma分布作为其先验.在平方损失下,利用Gibbs抽样和切片抽样给出了该模型参数的Bayes估计.最后,通过Monte Carlo模拟表明该Bayes估计是有效的.

英文摘要：

Abstract In this paper, we propose statistical analysis of Weibull distribution with Type-Ⅰ and Type-Ⅱ censored samples of the step-stress accelerated life testing under CE model. The likelihood function of the scale parameter with normal stress level is obtained using accelerated coefficient and acceleration equations to convert scale parameters under various accelerated stress levels into scale parameters under normal stress level. Scale parameters and accelerated coefficient are chosen as the conjugate prior and noninformative prior when choosing priors of parameters. Beta distribution and Gamma distribution are chosen as the priority when shape parameter m 〈 1 and m 〉 1, respectively. Bayesian estimation of this model under square loss is proposed using Gibbs sampling and Slice sampling. Fi-nally, statistical results show that the Bayesian estimation is efficient through Monte Carlo simulation.