中国安全科学学报 ›› 2024, Vol. 34 ›› Issue (4): 128-134.doi: 10.16265/j.cnki.issn1003-3033.2024.04.0839

• 安全工程技术 • 上一篇    下一篇

基于Apriori算法的氢安全事故统计分析

于溪芮(), 何旭, 孔得朋**()   

  1. 中国石油大学(华东) 机电工程学院,山东 青岛 266580
  • 收稿日期:2023-12-12 修回日期:2024-03-14 出版日期:2024-04-28
  • 通讯作者:
    **孔得朋(1985—),男,山东肥城人,博士,教授,主要从事油气及新能源的安全利用等方面的研究。E-mail:
  • 作者简介:

    于溪芮 (1998—),女,山东潍坊人,硕士研究生,研究方向为油气安全工程及应急。E-mail:

  • 基金资助:
    国家重点研发计划项目(2021YFB4000905)

Apriori algorithm-based statistical analysis of hydrogen accidents

YU Xirui(), HE Xu, KONG Depeng**()   

  1. College of Mechanical and Electrical Engineering, China University of Petroleum, Qingdao Shandong 266580, China
  • Received:2023-12-12 Revised:2024-03-14 Published:2024-04-28

摘要:

为探究氢安全事故不同特征因素间的相关性,以827起涉及氢气系统的安全事故为数据集,统计分析事故时间、应用阶段、点火源、事故致因及事故后果等特征因素。首先,构建事故致因的关联规则网络,讨论事故致因相关性;然后,交互分析事故致因和事故后果,明确事故因果关系;最后,基于Apriori算法挖掘氢安全事故的6类特征因素的关联规则。结果表明:氢安全事故中设备失效与设计错误、操作错误与意外反应等6个事故致因组合相关性最高;未点燃的氢气释放与压力释放装置、通风系统等8个因素强关联,氢火灾与螺栓、自燃等9个因素强关联,氢爆炸与制氢、电火花、压缩机等5个因素强关联。

关键词: Apriori算法, 关联规则, 氢安全事故, 统计分析, 事故致因, 交互分析

Abstract:

To explore the correlation between different characteristic factors of hydrogen accidents, 827 safety accidents involving hydrogen systems were used as the data set to statistically analyze the characteristic factors such as accident date, application stage, ignition source, accident causes and consequences. First, a network of association rules for accident causation was constructed to discuss accident causation relevance. Then, the causes and consequences of the accident were analyzed interactively. Finally, Apriori algorithm-based association rule mining was performed on six types of characteristic factors of hydrogen accidents. The results show that six accident causation combinations, such as equipment failure and design error, operational error and accidental reaction, are the most highly correlated among hydrogen safety accidents. Unignited hydrogen release is strongly associated with eight factors, like pressure relief devices and ventilation systems. Hydrogen fires are strongly associated with nine factors, like bolts and spontaneous combustion. And hydrogen explosions are strongly associated with five factors, like hydrogen production, electrical sparks, and compressors.

Key words: Apriori algorithm, association rule, hydrogen accidents, statistical analysis, accident causation, interaction analysis

中图分类号: