基于近似失效点的多源异构数据融合可靠性评估模型  

Multi-source heterogeneous data fusion model for reliability evaluation based on approximate failure points

作　　者：杨童博 魏政 周坤 YANG Tongbo;WEI Zheng;ZHOU Kun(Department of Statistics,School of Mathematics,Southwest Jiaotong University,Chengdu 611756,China;The Research Institute for Special Structures of Aeronautical Composite AVIC,Jinan 250023,China;Hefei General Machinery Research Institute Co.,Ltd.,Hefei 230031,China)

机构地区：[1]西南交通大学数学学院统计系,成都611756 [2]航空工业济南特种结构研究所,济南250023 [3]合肥通用机械研究院有限公司,合肥230031

出　　处：《机械强度》2025年第3期136-142,共7页Journal of Mechanical Strength

基　　金：国家自然科学基金项目(U1830133)。

摘　　要：为解决多源异构数据的融合问题,提高可靠性评估的精度,根据D-S(Dempster-Shafer)理论和最小二乘法相结合的思想,提出一种基于近似失效点的多源异构数据融合方法。首先,针对单一数据源构造概率分配并进行加权融合得到概率包络,建立基于近似失效点的分布拟合模型。其次,采用最小二乘法得到参数估计值并构造面积度量指标确定真实的失效分布函数,进而完成可靠性评估。最后,通过算例验证了该方法的可行性与有效性,相较于Bayes方法精度更高。In order to solve the problem of multi-source heterogeneous data fusion and improve the accuracy of reliability evaluation,a multi-source heterogeneous data fusion method based on approximate failure point was proposed by using D-S theory and the least squares method.Firstly,the probability envelope curves were obtained through constructing probability assignment for a single source of data and performing weighted fusion.A distribution fitting model based on approximate failure points was also established.Secondly,the parameter estimation value was obtained by the least squares method,and the area metric was constructed to determine the true failure distribution function,and then the reliability assessment was completed.Finally,the feasibility and effectiveness of the proposed method were verified by examples,and the accuracy was higher than that of Bayes method.

关 键 词：多源异构数据 D-S证据理论 近似失效点 最小二乘估计 面积度量指标 

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