在一般逐步I型区间截尾情形下,研究广义指数产品寿命试验的统计分析与优化设计问题。基于极大似然理论,利用EM算法给出参数的极大似然估计(MLEs)及可靠性指标的统计推断。依据缺损信息原则计算Fisher信息阵,据此确定最优截尾方案。采用不同的方案对估计结果进行模拟比较,从而得出受试产品的最优分组数及最优观测时刻。最后,应用算例验证方法的有效性。
This paper discusses the statistics inference and optimal plan for the generalized exponential components, when the sample are general progressively Type-I interval censored. Based on Maximum likelihood theory, we obtain Maximum likelihood estimations (MLEs) and estimations of reliability index for unknown parameters by EM algorithm. Fisher information matrixes axe computed using the missing value principle, accordingly the optimal progressive censoring plans axe determined. In order to obtain the optimal plan, Monte~Caxlo simulation results axe compared under different testing schemes. Finally, a case is presented to illustrate the validity of the method.