两类高度截尾数据及其参数估计问题  

Two Types of Heavily Censored Data and Associated Parameter Estimation Problem

作　　者：蒋仁言[1] JIANG Renyan(Faculty of Automotive and Mechanical Engineering,Changsha University of Science and Technology,Changsha 410114)

机构地区：[1]长沙理工大学汽车与机械工程学院,长沙410114

出　　处：《机械工程学报》2023年第10期374-382,共9页Journal of Mechanical Engineering

基　　金：国家自然科学基金资助项目(71771029)。

摘　　要：可靠性数据分析常常会遇到两类高度截尾数据。一类是高度随机截尾数据,另一类是由预防维修活动产生的截尾数据。如何基于这两类数据建立可靠性模型是一个极具挑战性的问题。通过数值仿真和对比分析,介绍高度随机截尾数据的生成背景和最新的参数估计方法,例证经典方法的不适用性和最新方法的有效性。针对预防维修截尾数据,指出现有研究的不足,着重分析两种典型的数据生成机制,提出采用独立竞争风险模型、比例剩余寿命模型和混合假定模型进行参数估计的新方法,通过采用仿真数例和实例充分例证了选择恰当的建模假定的极端重要性。主要的结论是现有面向高度随机截尾数据所开发的参数估计方法是相对地精确的;所提出的面向预防维修截尾数据的参数估计方法是有效的,估计方法的选择依赖于对数据生成机制的理解或数据特征分析。Reliability data analysis often faces two types of heavily censored data.One type of data is randomly censored and the other type of data is generated by preventive maintenance actions.It is challenging to construct reliability models using these two types of censored data.This research introduces the mechanism of generating heavily censored data and recently proposed parameter estimation methods through numerical simulation analysis.Results of numerical simulation show that recently proposed parameter estimation approaches are more effective than the classical methods when processing heavily censored data.After that,research gaps of processing preventive maintenance censored data are identified.Simulation is used to analyze mechanisms of generating two typical preventive maintenance censored data.Three new parameter estimation methods based on the independent competing risk model,the proportional residual life model,and the mixed assumption model are then developed.Both simulation and real-world datasets are analyzed to validate the importance of selecting appropriate assumption of modelling.Analysis results show that parameter estimation methods developed for highly randomly censored data can deliver relatively accurate estimates;the parameter estimation methods for preventive maintenance censored data are efficient,and the estimation method selection depends on the data generation mechanism or data characteristics analysis.

关 键 词：寿命数据分析 高度截尾数据 随机截尾 预防维修截尾 参数估计 

分 类 号：TB114[理学—概率论与数理统计]