基于知识图谱关系路径推理及相似度判别的机械故障源精确定位方法  

Accurate location method of mechanical fault source based on knowledge graph relationship path inference and similarity discrimination

作　　者：奥帅 王嘉欢 马波[2] AO Shuai;WANG Jiahuan;MA Bo(Sandaogou Coal Mine,Shaanxi Deyuan Fugu Energy Co.,Ltd.,CHN Energy Group.,Yulin 719407,China;College of Mechanical and Electrical Engineering,Beijing University of Chemical Technology,Beijing 100029,China)

机构地区：[1]国家能源集团陕西德源府谷能源有限公司三道沟煤矿,陕西榆林719407 [2]北京化工大学机电工程学院,北京100029

出　　处：《机电工程》2025年第2期387-396,共10页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(62273025)。

摘　　要：应用知识图谱进行机械故障诊断时,存在难以精确定位故障根源的问题,为此,提出了一种基于知识图谱关系路径推理及相似度判别的机械故障源精确定位方法。首先,分析了现有的基于知识图谱的机械故障诊断研究现状,并详细阐述了知识图谱的构建方法、基于关系路径的推理方法、基于关系权重的相似度计算方法,以及基于德尔菲法的权重设置方法;然后,采用“自顶向下”的方式构建了机械设备故障诊断知识图谱,并对基于关系路径的推理方法进行了优化设计,在正向推理得到可能的故障原因基础上,进行了反向推理,得到了与可能故障原因相关联的故障现象集合;最后,对输入的故障现象集合与关联故障现象集合进行了相似度判别,并从大到小进行了排序,将排序好的故障原因作为最终的诊断结论,对双滚筒采煤机的典型故障进行了诊断,构建了采煤机的诊断知识图谱,验证了该方法的有效性。研究结果表明:利用知识图谱关系路径推理及相似度判别方法,针对故障案例“截割电机机身温度偏高,截割电机阻值低于正常范围”,诊断得出截割电机绕组绝缘失效的故障概率为inf(无穷大),截割电机风扇故障的概率为0.735 2,采煤机过负荷运行的故障概率为0.548 8,与故障机理相符,实现了机械故障源精确定位目的。该方法可为改进基于知识图谱的机械故障诊断方法提供参考。Accurate location method of mechanical fault source based on knowledge graph relational path reasoning and similarity discrimination was proposed,to address the challenge of locating the root cause of mechanical faults using knowledge graphs.First,the existing research on mechanical fault diagnosis based on knowledge graphs was analyzed.The construction method of knowledge graphs,the relational path-based reasoning method,the similarity calculation method based on relational weight,and the weight setting method using the Delphi method were detailed.Then,a knowledge graph for mechanical equipment fault diagnosis was constructed in a top-down approach,and the reasoning method based on relational paths was optimized.Using forward reasoning,possible fault causes were identified.Reverse reasoning was then applied to obtain the set of fault phenomena associated with these possible causes.Finally,the similarity between the input fault phenomenon set and the associated fault phenomenon set was determined,and the sorted fault causes were taken as the final diagnosis conclusion.The typical faults of the double drum shearer were diagnosed,the diagnosis knowledge graph of the shearer was constructed,and the effectiveness of the method was verified.The research results show that by using the knowledge graph relationship path reasoning and similarity discrimination method,for the fault case of "the body temperature of the cutting motor is on the high side and the resistance value of the cutting motor is lower than the normal range",the fault probability of cutting motor winding insulation failure is diagnosed as inf(infinity),the fault probability of cutting motor fan failure is 0.735 2,and the fault probability of shearer overload operation is 0.548 8.These results are consistent with fault mechanisms and achieve the goal of precise localization of mechamical fault sources.This method can provide a reference for the improvement of mechanical fault diagnosis methods based on knowledge graphs.

关 键 词：采煤机械 机械故障诊断 知识图谱关系路径推理 相似度判别 机械设备故障诊断知识图谱 故障精确定位 

分 类 号：TH17[机械工程—机械制造及自动化] TD421[矿业工程—矿山机电]