分散性和落入概率整体推断方法  

INTEGRAL INFERENCE METHOD FOR SCATTER AND PROBABILITY

作　　者：傅惠民[1] 

机构地区：[1]北京航空航天大学小样本技术研究中心,北京100083

出　　处：《机械强度》2006年第2期216-219,共4页Journal of Mechanical Strength

基　　金：国防科技预研项目(413200204)资助。~~

摘　　要：针对正态分布均值为零或已知(包括均值能被正确仿真)的情况,提出一种分散性和落入概率整体推断方法。该方法通过仿真结果与试验数据的结合,将多个不同状态下的试验数据作为一个整体进行统计分析,给出正态分布方差和落入概率的整体估计与置信区间估计。与传统方法相比,该方法不但具有更高的精度,而且解决了一种状态下只有一个试验数据时分散性和落入概率评定的难题。文中还给出一种方差仿真结果的极小子样检验方法。An integral inference method for scatter and probability of normal distribution with zero mean is presented. By combining the results of simulation and the experiment data, the method can totally infer the experiment data of normal distributions in multistate. Fm＇thermore, it can establish the variance and the probability estimators of normal distribution, and give their confidence limits and intervals in each state. Compared with traditional methods, this method not only has higher precision but also solves the problem of scatter and probability assessment with only one experiment datum of normal distribution in a state. A test method for simulation results of variances in multi-state with very small sample is also presented.

关 键 词：正态分布 分散性 落入概率 整体估计 小于样 可靠性 仿真 

分 类 号：O212.1[理学—概率论与数理统计] TB114.3[理学—数学]