Weibull和正态分布不完全数据可靠性评估方法  

Reliability Assessment for Weibull and Normal Distributions with Incomplete Data

作　　者：傅惠民[1] 郭建超 李子昂 FU Hui-min;GUO Jian-chao;LI Zi-ang(Research Center of Small Sample Technology,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学小样本技术研究中心,北京100191

出　　处：《机电产品开发与创新》2024年第5期1-6,共6页Development & Innovation of Machinery & Electrical Products

基　　金：国家自然科学基金(U2037602)《月基装备自主操控与多机协同基础理论与关键技术研究》。

摘　　要：提出一种两参数Weibull分布和正态分布定数截尾数据可靠性评估方法,建立了其高置信度下的可靠寿命和可靠度单侧置信限计算公式。同时,给出一种能够充分开发利用以往试验数据,并与当前试验数据有机融合进行可靠性评估的方法,由于增大了信息量,从而可以显著提高当前产品可靠性评估精度。在此基础上,还进一步将上述方法推广用于两参数Weibull分布和正态分布的定时截尾数据、无失效数据以及一般不完全数据的可靠性评估,从而实现了不完全数据情况机电产品高精度小样本可靠性评估。与传统的需查表计算的BLUE和BLIE等方法相比,本文方法不但理论上更加严谨,而且评估精度更高,工程计算也更加便捷。Reliability assessment methods for two-parameter Weibull and normal distributions with type-Ⅱ censored data are proposed,enabling high-confidence inference of the one-sided confidence limit for reliable life and reliability.Moreover,a method that can make full use of past test data and integrate it with current test data for reliability assessment is presented,which can significantly improve the accuracy of current product reliability assessments due to the increased information.Based on this,the above methods are further extended to the reliability assessment of two-parameter Weibull and normal distributions with type-Ⅰ censored data,zero-failure data,and general incomplete data.This extension achieves a high-precision and small-sample reliability assessment of electromechanical products under incomplete data conditions.Compared to the traditional methods such as BLUE and BLIE,which require complex table querying,the proposed method is not only more theoretically rigorous but also offers higher assessment accuracy and simpler calculation steps.

关 键 词：定数截尾数据 定时截尾数据 不完全数据 可靠性评估 小样本 置信限 

分 类 号：O213.2[理学—概率论与数理统计] TB114.3[理学—数学]