Control and cause analysis for high-risk operations combined knowledge graph and Apriori algorithm

doi:10.16265/j.cnki.issn1003-3033.2026.04.0702

Abstract

Abstract:

In order to improve the safety management level of high-risk operations, an intelligent management and control method was proposed based on the knowledge graph and the Apriori algorithm. First, some high-risk operation accidents were collected. Then, the accident causes were extracted and classified automatically by Bidirectional Encoder Representations from Transformers-Bidirectional Long Short-Term Memory-Conditional Random Field(BERT-BiLSTM-CRF)model, which constructed a knowledge graph of high-risk operation accidents, storing relevant accident data. Subsequently, an index system of high-risk operation causative factors was established, combined with the corresponding operation standard specification. Furthermore, the correlation among causative factors was determined by the Apriori algorithm. Finally, some countermeasures were proposed for targeted control. In this paper, the application of confined space operations was used to demonstrate this method. The results showed that the main causes of confined space accidents are not wearing safety protective equipment", blind rescue", not being equipped with gas detection and other equipment", missing safety warning signs", insufficient safety education", and inadequate safety management". Moreover, the correlation between insufficient safety education" and not wearing safety protective equipment" is greater (80.7% support, 60.41% confidence), and the relationship between insufficient safety education" and blind rescue" also has a strong association (80.7% support, 55.88% confidence).

Key words: knowledge graph, Apriori algorithm, high-risk operation, intelligent management, causative factor

CLC Number:

X928.03事故预防与预测

Guo Hanjun, Cui Huaying, Hu Zhenqi, Kang Rongxue, Zhao Jinlong. Control and cause analysis for high-risk operations combined knowledge graph and Apriori algorithm[J]. China Safety Science Journal, 2026, 36(4): 75-84.

Figures/Tables 17

Table 1

Fig.1

Fig.2

Fig.3

Fig.4

Table 2

Fig.5

Fig.6

Fig.7

Fig.8

Table 3

Table 4

Fig.9

Table 5

Index system of causative factors for accidents in some confined space operations

一级指标	二级指标	三级指标
人员因素	操作规程执行错误	未对现场进行通风检测;未对现场有害气体进行检测; 违章操作;未经允许作业;未办理作业审批手续; 未对设备进行安全检查;未检测作业现场环境
	自身行为(动作、姿势、站位规范性要求)不规范	站在危险区域作业;未按正确方式作业;作业不彻底
	未察觉生产现场状态条件存在安全风险	未及时发现现场环境变化;未能检测出气体浓度超标
	未佩戴个体防护装备或佩戴不正确	未佩戴安全防护装备;防护装备佩戴不正确
	应急处置不当	盲目施救;应急救援不当
	技能状况	安全意识薄弱;技能不足
设备因素	设备系统设计与选择的完整性与工作适配性	设备设计不符合实际要求;设备位置、改装设置不合理;使用的设备与现场不匹配
	设备设施质量完好	安全防护装置失效;设备本身存在缺陷;安全防护装置缺失
	作业前生产场所的安全状态与措施	未配备气体检测、通风等设备;管道阀门未关闭;管道封堵;设备之间无连锁
	生产作业中出现的突发异常不安全因素	零部件自燃;设备故障;设备电源未切断
环境因素	物理环境	照明不足;温度较高
	空间环境	封闭空间;作业空间狭窄;有害气体超标;缺氧环境;通风不彻底
	安全防护	安全警示标志缺失;现场无安全防护措施;安全距离不足;未设置安全警戒区
	作业干扰因素	交叉作业;有其他干扰的杂物
管理因素	技术管理	未制定各项规章制度、操作规程、处置方案等文件;安全生产规章制度、操作规程等文件不健全;安全措施内容制定不全面
	生产组织管理	劳动用品配备不足;未安排专人进行现场安全管理; 防护设备设施配备不足;检测设备配备不足
	作业前准备管理	风险辨识不到位;隐患排查治理不到位;安全技术交底不到位;设备、环境等安全检查不到位
	作业过程管理	安全管理不到位;安全生产规章制度和安全操作规程落实不到位
	基础管理	安全教育培训不到位;未建立安全生产责任制或安全主体责任落实不到位;未制定有限空间作业安全管理制度;安全投入不足
	合规管理	人员无证上岗;外包作业;无证经营;作业审批不到位

Table 5

Fig.10

Fig.11

Fig.12

References 16

[1]	刘辉, 郑向莞, 张智超, 等. 定量改进JHA法的建筑施工高危作业风险分析[J]. 中国安全科学学报, 2019, 29(4):146-151. doi: 10.16265/j.cnki.issn1003-3033.2019.04.023
	Liu Hui, Zheng Xiangwan, Zhang Zhichao, et al. Risk analysis of high-risk operations in construction by a quantitatively improved JHA method[J]. China Safety Science Journal, 2019, 29(4):146-151. doi: 10.16265/j.cnki.issn1003-3033.2019.04.023
[2]	章博, 刘悦, 姚明, 等. 基于JSA方法的高风险作业在线实践平台[J]. 安全与环境学报, 2019, 19(2):509-513.
	Zhang Bo, Liu Yue, Yao Ming, et al. Practice teaching platform for high risk operation based on the JSA method[J]. Journal of Safety and Environment, 2019, 19(2):509-513.
[3]	Peres S C, Hendricks J W. A systems model of procedures in high-risk work environments: empirical evidence for the safety model 2 approach using the interactive behavior triad[J]. Safety Science, 2024, 172: DOI: 10.1016/j.ssci.2023.106328.
[4]	Jiang Weiguang, Liu Yuhan, Chen Ke, et al. Early-warning of unsafe hoisting operations: an integration of digital twin and knowledge graph[J]. Developments in the Built Environment, 2024, 19: DOI: 10.1016/j.dibe.2024.100490.
[5]	Zhou Zhipeng, Yu Xinhui, Magoua J J, et al. Integrating machine learning and a large language model to construct a domain knowledge graph for reducing the risk of fall-from-height accidents[J]. Accident Analysis & Prevention, 2025, 215: DOI: 10.1016/j.aap.2025.108009.
[6]	田晓敏, 李晓冬. 基于Apriori算法和贝叶斯网络的高处坠落事故致因分析[J]. 重庆科技大学学报(自然科学版), 2024, 26(6):80-87.
	Tian Xiaomin, Li Xiaodong. Analysis of falling accident causes based on Apriori algorithm and Bayesian network[J]. Journal of Chongqing University of Science and Technology(Natural Science Edition), 2024, 26(6):80-87.
[7]	国汉君. 内-外因事故致因理论与实现安全生产的途径[J]. 中国安全科学学报, 2007, 17(7):46-53,179.
	Guo Hanjun. Theory of Accident causation based on internal-external factors and approaches for work safety[J]. China Safety Science Journal, 2007, 17(7): 46-53,179.
[8]	Li Wei, Du Yajun, Li Xianyong, et al. UD_BBC: named entity recognition in social network combined BERT-BiLSTM-CRF with active learning[J]. Engineering Applications of Artificial Intelligence, 2022, 116: DOI: 10.1016/j.engappai.2022.105460.
[9]	Yu Yuqing, Wang Yuzhu, Mu Jingqin, et al. Chinese mineral named entity recognition based on BERT model[J]. Expert Systems with Applications, 2022, 206: DOI: 10.1016/j.eswa.2022.117727.
[10]	Witten I H, Frank E, Hall M A, et al. Practical machine learning tools and techniques[J]. Data Mining, 2005, 2(4): 403-413.
[11]	Xu Ruihua, Luo Fan. Risk prediction and early warning for air traffic controllers' unsafe acts using association rule mining and random forest[J]. Safety Science, 2021, 135: DOI: 10.1016/j.ssci.2020.105125.
[12]	况宇琦. 基于数据挖掘的塔吊事故案例分析研究[D]. 武汉: 华中科技大学, 2021.
	Kuang Yuqi. Research on case analysis of tower crane accidents based on data mining[D]. Wuhan: Huazhong University of Science and Technology, 2021.
[13]	DB11T 1584—2018 有限空间中毒和室息事故勘查作业规范[S].
	DB11T 1584-2018 Specification for the investigation of poisoning and asphyxia accidents in confined spaces[S].
[14]	DB33/T 707—2022 工贸企业受限空间作业安全技术规范[S].
	DB33/T 707-2022 Safety technical regulations for working in confined spaces of enterprises[S].
[15]	DB14/T 2124—2020 冶金工贸企业有限空间作业安全规范[S].
	DB14/T 2124-2020 Industrial and trade enterprises operating in confined space safety technical specifications.
[16]	王勇毅, 姜亢, 郭建中, 等. 职业中毒场所盲目施救行为的心理因素分析与对策研究[J]. 中国安全科学学报, 2012, 22(11):11-15.
	Wang Yongyi, Jiang Kang, Guo Jianzhong, et al. Analysis and countermeasures research on psychological factors in blind rescue behavior in occupational poisoning place[J]. China Safety Science Journal, 2012, 22(11):11-15.

范围	定义	事故类型
动火作业	在直接或间接产生明火的工艺设施以外的禁火区内从事可能产生火焰、火花、炽热表面的非常规作业	火灾、爆炸、中毒等
起重作业	通过起重机械实施重物垂直起升与水平移动的操作活动	物体打击、起重机倒塌、起重伤害等
有限空间作业	封闭或部分封闭,进出口较为狭窄有限,未被设计为固定工作场所,自然通风不良,易造成有毒有害、易燃易爆物质积聚或氧含量不足的空间	中毒、缺氧、燃爆等
高处作业	人在一定位置为基准的高处进行的作业	高处坠落、物体打击等

实体间的关系类型	实体间的关系
从属关系	事故名称,事故类型
从属关系	事故名称,事故原因(人员、设备、环境、管理)
因果关系	管理因素,人员因素
	管理因素,设备因素
	管理因素,环境因素

模型	精确率	召回率	F₁分数
BERT-BiLSTM-CRF模型	0.931 7	0.938 4	0.931 8
BiLSTM-CRF模型	0.926 1	0.934 7	0.922 7
IDCNN-CRF模型	0.919 9	0.931 9	0.919 1

事故名称	直接原因	间接原因	人员因素	设备因素	环境因素	管理因素	事故类型	事故后果
北京朝阳“7·30”较大中毒和窒息事故	井下缺氧,作业人员未测定气体含量、未通风、未配备呼吸器具和安全带	管理制度不落实、劳动用品管理混乱、培训不到位、安全检查不到位	未检测现场环境	—	氧气含量不足	劳动用品未配备	窒息死亡	3人死亡
			未佩戴呼吸器	—		培训不到位
			未佩戴安全带	—		管理不到位
			—	—		检查不到位