基于生存分析与极大似然法的可靠性评估  

Reliability estimation based on survival analysis and maximum likelihood

作　　者：戴怡[1] 刘琴[2] DAI Yi;LIU Qin(College of Mechanical Engineering, Tianjin University of Technology and Education, Tianjin 300222, China;College of Science,Tianjin University of Technology and Education, Tianjin 300222,China)

机构地区：[1]天津职业技术师范大学机械工程学院,天津300222 [2]天津职业技术师范大学理学院,天津300222

出　　处：《计算机集成制造系统》2020年第11期2976-2981,共6页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(50875186,11702193)。

摘　　要：为了对比生存分析与极大似然法在可靠性评估方面的优劣,并选择适合有替换定时截尾试验的评估方法,研究了基于生存分析与极大似然法的可靠性评估方法。由于认为有替换定时截尾模型更适合数控机床可靠性评估,推导了有替换定时截尾模型与随机删失模型之间的关系。采用蒙特卡洛方法进行研究,得出常规时间下(即截尾时间是平均寿命真值1.4~2.5倍)极大似然法评估结果优于生存分析方法的结论,并计算出有替换定时截尾试验的截尾时间与置信概率之间的相互关系。To contrast survival analysis with maximum likelihood in reliability estimation,and to choose better estimation method in timing censored test with replacement,the reliability estimation method based on survival analysis and maximum likelihood was studied.Because timing censored test with replacement had been considered to be most appropriate for reliability estimation of numerical control machine tool,the relationship between the model of randomly censored and timing censored test with replacement was deduced.With Monte Carlo method,the conclusion that maximum likelihood was better than survival analysis during common test time(the test time was 1.4~2.5 times that of mean time between failures)was obtained,and the relationship between the censored test time and confidence probability of timing censored test with replacement was calculated.

关 键 词：有替换定时截尾 随机截尾 生存分析 极大似然法 平均无故障工作时间 数控机床 可靠性评估 

分 类 号：TG659[金属学及工艺—金属切削加工及机床] O213.2[理学—概率论与数理统计]