基于证据推理的可靠度近似计算方法  

Evidence Theory-based Method and Algorithm for Calculation of Reliability

作　　者：郭惠昕[1] 胡冠昱[1] 谈锋[1] 

机构地区：[1]长沙大学机电工程系

出　　处：《农业机械学报》2008年第5期128-132,共5页Transactions of the Chinese Society for Agricultural Machinery

基　　金：湖南省教育厅高等学校科学研究重点资助项目(项目编号:03A031)

摘　　要：提出了一种基于证据推理的可靠度近似计算方法。定义结构失效的命题为F,通过证据推理求出F的信度Bel(F)和似真度pl(F),Bel(F)和pl(F)的均值可作为失效概率的近似值。研究了该方法的数值计算方法,给出了具体的计算步骤,并分析了影响计算精度与计算效率的关键因素。研究表明:随机变量分布区间的等分数k越大则失效概率的计算精度越高;而随机变量个数n和k越大,计算效率则越低。开发了用该方法计算可靠度的专用程序,并通过应用实例验证了该方法的正确性和有效性。A method of approximate reliability calculation was introduced based on evidence reasoning. F was defined as an event of structure failure. The belief measure of F （Bel（F）） and the plausibility measure of F （pl（F）） were calculated using evidence theory. [Bel（F）, pl（F） ] was used as the interval estimation of failure probability. The distributing domains of n random variables in performance function of engineering structure were respectively split into k intervals. The average of Bel（ F ） and pl （F） could be considered as the approximate estimation of failure probability. A numerical algorithm for realizing the introduced method was investigated and its calculation steps were presented, The key influencing factors on the evaluation precision and calculation efficiency were illuminated subsequently. The greater the k is, the more precise the calculated failure probability is, and the calculation efficiency decreases with the increasing of k and n. Finally, a program for realizing this algorithm was developed and two examples were presented to demonstrate the validity of the introduced method.

关 键 词：可靠度 失效概率 计算方法 证据理论 

分 类 号：TB114.3[理学—概率论与数理统计]