Computer Engineering and Applications ›› 2024, Vol. 60 ›› Issue (16): 311-318.DOI: 10.3778/j.issn.1002-8331.2305-0459

• Engineering and Applications • Previous Articles     Next Articles

Research on Fault Entity Relation Extraction of Mine Hoisting System

DANG Xiaochao, YE Hanxin, DONG Xiaohui, LI Fenfang, ZHU Zhongyan   

  1. 1.College of Computer Science & Engineering, Northwest Normal University, Lanzhou 730070, China
    2.Longshou Mine, Jinchuan Group Co., Ltd., Jinchang, Gansu 737103, China
  • Online:2024-08-15 Published:2024-08-15

矿井提升系统的故障实体关系抽取研究

党小超,叶汉鑫,董晓辉,李芬芳,祝忠彦   

  1. 1.西北师范大学 计算机科学与工程学院,兰州 730070
    2.金川集团股份有限公司 龙首矿,甘肃 金昌 737103

Abstract: Mine hoisting system is the core equipment of mining production, most of the important data such as fault maintenance records of mine hoisting system are stored in text form, and the construction of mine hoisting system fault graph can make effective use of these data. Relationship extraction is an important part of fault graph construction. In order to improve the accuracy of the relationship extraction in constructing the fault graph of mine hoisting system, this paper proposes a model fusion method for relationship extraction. The method uses natural language processing techniques to extract relevant fault descriptions and machine component information from maintenance records, and uses a self-built model to extract fault entity relationships from the textual data. To address the problem of the lack of fault text data sets for hoisting systems, this paper collects maintenance records data and existing fault literature of a large non-ferrous metal mining enterprise in northwest China, cleans and organizes these data, and establishes a fault text data set for mine hoisting systems. Experimental results on the self-built data set show that the proposed model has higher accuracy compared to traditional models.

Key words: mine hoisting system, knowledge graph, relationship extraction, self-built data set

摘要: 矿井提升系统是矿业生产的核心设备,矿井提升系统的故障维护记录等重要资料多数以文本形式存储,构建矿井提升系统故障图谱可以有效利用这部分资料。关系抽取是故障图谱构建的重要环节,为了提高矿井提升系统故障图谱构建的关系抽取环节的准确率,提出一种模型融合的关系抽取方法。该方法使用自然语言处理技术从维修点检记录中提取出相关的故障描述和机器部件信息,利用自建模型从这些文本数据中提取出故障实体关系。针对目前提升系统故障文本数据集缺乏的问题,收集了西北某大型有色金属矿产企业的点检维修记录数据和现有的矿井提升系统故障文献,将这些数据进行清洗和整理,建立了一个矿井提升系统故障文本数据集。将该模型与传统模型在自建数据集上进行实验,实验结果证明该模型相较于传统模型有更高的准确率。

关键词: 矿井提升系统, 知识图谱, 关系抽取, 自建数据集