串联系统小样本可靠度直接评估方法  

Direct Assessment Method for Series System Reliability Based on Small Sample Data

作　　者：傅惠民[1] 李子昂 郭建超 FU Hui-min;LI Zi-ang;GUO Jian-chao(Research Center of Small Sample Technology,Beihang University,Beijing100191,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4期1-5,共5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：建立了一种串联系统小样本可靠度直接评估方法,能够根据子系统的可靠度置信限直接给出串联系统的可靠度置信限,解决了目前常用的LM法和MML法等传统方法需将信息量较大的连续型分布的子系统试验数据转换为信息量较小的成败型数据,造成试验信息大量丢失,导致可靠度评估精度不高或偏于危险的问题。在此基础上,针对工程上许多成败型试验转计量型试验的机构都是正态分布串联系统的情况,建立了正态分布串联系统小样本可靠度直接评估方法,该方法能够根据子系统的小样本计量型数据对串联系统可靠度进行高精度评估。大量的Monte Carlo模拟验证和实例计算结果表明,传统方法在小样本情况下误差较大,甚至无法满足置信度要求,本文方法则不仅能够满足置信度要求,而且评估精度更高。此外,本文方法对工程上常用的指数分布、 Weibull分布和正态分布等连续分布均适用。A direct assessment method for series system reliability using small-sample data is proposed in this paper,which can assess the reliability confidence limit of a series system based on the reliability confidence limits of its subsystems.This approach overcomes the limitations of traditional methods such as the LM and MML,which require the conversion of continuous subsystem data into less informative binomial data.This conversion often results in low accuracy or a dangerous estimate in reliability assessment.On this basis,a reliability assessment method for a series system with normal-distributed subsystems is established.This method is capable of assessing the series system reliability based on the small-sample metrological data from the subsystems.Extensive Monte Carlo simulations and examples demonstrate that the proposed method not only meets the confidence requirement but also offers higher assessment accuracy compared to traditional methods.Furthermore,the proposed methodology applies to common continuous distributions,including exponential,Weibull,and normal distributions.

关 键 词：串联系统 小样本 可靠度评估 置信限 成败型试验 计量型试验 

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