摘要
针对工程上常见的加速寿命试验中加速系数未知的情况,建立一种小样本加速寿命试验方法,该方法只需在两个加速应力水平下进行加速寿命试验,即可对产品在正常使用应力水平下的寿命和可靠性进行评估。根据正常使用条件下的可靠度和置信水平要求,采用两个加速应力水平下具有相同可靠度和置信水平的寿命单侧置信下限之比作为加速系数,该加速系数考虑了可靠度和置信水平的影响,在推断正常应力水平下的可靠寿命和可靠度单侧置信下限时,可以有效减小误差,提高精度。文中详细讨论了阿伦尼斯模型、逆幂律模型和指数模型等三种常用加速模型下的加速系数、寿命预测和可靠性评估。大量工程算例和Monte Carlo模拟验证表明,传统方法需在多个加速应力水平下进行寿命试验,与之相比,本文方法可以节省大量试样和试验时间。
For the common situation in accelerated life test(ALT)that acceleration factor is unknown,a novel ALT method with small sample is proposed,which can evaluate the lifetime and reliability at normal stress level by the test data under only two accelerated stress levels.According to the requirement of reliability and confidence level under normal use condition,the acceleration factor is defined as the ratio of one-side lower confidence limits of reliable life with the same index requirement under accelerated stress levels.This acceleration factor can take both reliability and confidence level into consideration,enabling to reduce error and improve accuracy when extrapolating the one-side lower confidence limits of reliable life and reliability at normal stress level.For the frequently used accelerated models of Arrhenius model,inverse power model and exponential model,corresponding acceleration factor,lifetime prediction and reliability assessment are discussed in detail.Calculating results of engineering examples and Monte Carlo simulation indicate that,compared with the conventional method which needs to conduct the test under multiple stress levels,our proposed method can save a large quantity of test specimens and time.
作者
傅惠民
文歆磊
吴琼
FU Hui-Min;WEN Xin-Lei;WU Qiong(Research Center of Small Sample Technology,Beihang University,Beijing100191,China;Institute of Spacecraft System Engineering,Beijing 100094,China)
出处
《机电产品开发与创新》
2020年第3期1-4,共4页
Development & Innovation of Machinery & Electrical Products
基金
2018年智能制造综合标准化项目《基于数字仿真的机械产品可靠性测试方法标准研究与试验验证》
国家重点基础研究发展计划(2012CB720000)。
关键词
加速寿命试验
加速系数
可靠性评估
寿命预测
小样本
置信水平
置信下限
accelerated life test
acceleration factor
reliability assessment
lifetime prediction
small sample
confidence level
lower confidence limit
