Abstract: Equipment spot-inspection records are essential information sources supporting fault reason analysis.However,there is a lack of effective mining of root cause information in equipment spot-inspection failures to improve the reliability of preventive equipment maintenance.Therefore,the causal theory was introduced into the manufacturing field for the first time to propose a causal knowledge modeling for root cause analysis of equipment spot-inspection failure.The event knowledge was extracted from the equipment spot-inspection fault documents to construct a fault operation and maintenance causal knowledge graph.The fault operation and maintenance causal knowledge rules were defined to establish a structural causal graph model.Then,an Intervention Sum Product Networks (ISPN)-based causal effect estimation learning model was designed to calculate the confounding influences in the fault knowledge and determine the main factors affecting equipment failures.Moreover,the implicit causal semantic links between nodes were complemented.The proposed method was tested with knowledge of metallurgical equipment spot inspection failure documentation.The feasibility of the causal knowledge model in estimating the causal effect among knowledge of the root causes for equipment failures was verified.

Key words: industrial knowledge graph, causal knowledge modeling, equipment spot-inspection faulty, root cause analysis

摘要： 设备点检记录是支撑故障原因分析与处理的重要信息来源,目前亟需对设备点检故障中的根因信息进行有效挖掘,以提升设备预防性维护的可靠性。鉴于此,首次将因果科学论引入制造领域,提出一种面向设备点检故障根因分析的因果知识建模方法。首先,从设备点检故障文档中提取事件知识,构建故障运维因果知识图谱;其次,定义故障运维因果知识规则,形成结构因果图模型;进而,设计一种基于ISPN的因果效应估计学习模型,对故障知识中混杂影响因素进行估计计算,挖掘出影响设备故障发生的语义关系,补全图谱节点间隐含的因果性语义链路;最后,以冶金设备点检故障文档的知识测试了所提方法,验证了因果知识模型估计设备故障根因知识间因果效应的可行性。

关键词: 工业知识图谱, 因果知识建模, 设备点检故障, 根因分析