面向多源信息融合的测试性建模与评估方法  

Testability modeling and evaluation method for multisource information fusion

作　　者：张西山 连光耀 布树辉[2] 李会杰 宋秦松 ZHANG Xishan;LIAN Guangyao;BU Shuhui;LI Huijie;SONG Qinsong(Chinese People’s Liberation Army 32181,Xi’an 710032,China;School of Aeronautics,Northwestern Polytechnical University,Xi’an 710072,China)

机构地区：[1]中国人民解放军32181部队,西安710032 [2]西北工业大学航空学院,西安710072

出　　处：《航空动力学报》2025年第1期416-426,共11页Journal of Aerospace Power

基　　金：装备预研领域基金(61400040302)。

摘　　要：针对由于信息缺乏难以进行装备测试性建模与评估的问题,提出了面向多源信息融合的测试性建模与评估方法。首先,结合测试性结构模型和贝叶斯网络模型,构建了包含系统结构层次、故障模式、测试、有向边和节点条件概率等元素组成的分层混合测试性模型;然后,提出了基于专家经验和试验数据相结合的测试性模型节点条件概率确定方法,建立了测试性评估推理模型,实现了小子样条件下测试性信息的准确描述;最后,通过实例分析验证了基于分层混合模型仿真的测试性评估结果的精确性,相比传统建模评估方法故障检测率降低0.79%,避免了传统方法的冒进。For the difficulty to carry out testability modeling and evaluation due to the lack of information,the testability modeling and evaluation method for multisource information fusion were studied.Firstly,referring to testability structure model and Bayesian networks model,a layered hybrid testability model consisting of system component unit,failure,mode,test,directed edge and node conditional probability was constructed.Then,the method of determining the node conditional probability was proposed by combining expert experience and experiment data,the testability evaluation reasoning model was established,and accurate description of probability information under small sample condition was realized;Lastly,it was verified that the testability evaluation result based on hierarchical hybrid model simulation has the highest accuracy,compared with the traditional modeling evaluation method,the fault detection rate was reduced by 0.79%.

关 键 词：信息融合 分层混合 测试性建模 测试性评估 小子样 

分 类 号：V19[航空宇航科学与技术—人机与环境工程] TP206[自动化与计算机技术—检测技术与自动化装置]