中国安全科学学报 ›› 2023, Vol. 33 ›› Issue (2): 9-15.doi: 10.16265/j.cnki.issn1003-3033.2023.02.0949

• 安全社会科学与安全管理 • 上一篇    下一篇

煤矿事故因素的自组织映射分布研究

张江石(), 李泳暾**(), 秦芳, 王慧超, 潘雨, 王梓伊   

  1. 中国矿业大学(北京) 应急管理与安全工程学院,北京 100083
  • 收稿日期:2022-09-19 修回日期:2022-12-15 出版日期:2023-02-28
  • 通讯作者:
    **李泳暾(1995—),男,山西长治人,博士研究生,主要研究方向为智能化安全管理、事故文本挖掘、事故数据挖掘等。E-mail:
  • 作者简介:

    张江石 (1973—),男,河南三门峡人,博士,教授,主要从事安全管理和煤矿粉尘方面的研究。E-mail:

Study on self-organizing mapping distribution of coal mine accident factors

ZHANG Jiangshi(), LI Yongtun**(), QIN Fang, WANG Huichao, PAN Yu, WANG Ziyi   

  1. School of Emergency Management and Safety Engineering, China University of Mining and Technology-Beijing, Beijing 100083, China
  • Received:2022-09-19 Revised:2022-12-15 Published:2023-02-28

摘要:

为进一步明确煤矿事故致因因素间的复杂耦合机制,引入自组织映射(SOM)神经网络。首先利用24Model分析煤矿事故因素,构建事故属性数据集;其次通过SOM算法,可视化分析因素与事故间的耦合关系,最后结合K-means聚类算法,研究煤矿事故致因中单因素及多因素耦合对事故的映射特征,并计算因素间的耦合程度。结果表明:“事故—单因素”映射分布可反映出单因素对事故影响程度的差异性,其中管理落实不到位、安全培训不到位以及制度文件不健全等因素对煤矿事故的影响较大;“事故—多因素耦合”映射分布得到直接因素类别中违章指挥、违规作业、操作失误和不安全物态4种因素的耦合作用较强;因素耦合度分析中“体系文件不健全、安全文化欠缺—人员组织不合理”、“配套设施不齐全—安全习惯不佳”等多对因素的耦合度大于0.8,存在较强的耦合关系和因果性。

关键词: 煤矿事故, 自组织映射(SOM), 事故致因因素, 耦合关系, 可视化

Abstract:

In order to further clarify the complicated coupling mechanism between the coal mine accident causation factor, SOM neural network was introduced to study the coupling relationship between factors. Firstly, 24Model was used to analyze the factors of coal mine accidents and the accident attribute dataset was built. The SOM algorithm was then used to visualize the coupling relationship between factors and accidents. Combined with the K-means clustering algorithm, the mapping characteristics of single-factor and multi-factor coupling to coal mine accident causes were studied, and the correlation degree between the factors was analyzed. The results show that the difference in impact degree of a single factor on the accident is reflected by "accident-single factor" mapping distribution. Factors such as inadequate management, inadequate safety training and imperfect institutional documents contributed more to the accident. Four factors, including illegal command, illegal operation, operation error and unsafe object state, are the strongest coupling effect in the "accident-multifactor coupling" mapping. There is a strong correlation with coupling degree≥0.8 between the factors such as "imperfect system documents, insufficient safety culture--unreasonable personnel organization" and "incomplete supporting facilities--poor habits" in the factor correlation analysis.

Key words: coal mine accident, self-organizing mapping(SOM), cause of accident factor, coupling relationship, visualization