基于图论的生产安全事故风险预警指标筛选及分析*

为提高事故风险预警准确性，降低事故发生频率，通过对现有生产安全事故案例进行因果连锁理论分析，构建事故成因结构图，采用图论中节点属性分析法，筛选事故风险预警指标，并利用最短路径介数和紧密度改进复杂网络概念，从局部与整体角度探讨事故风险预警指标敏感性。结果表明:人与物类风险预警指标敏感性较高且相近，环境类风险预警指标敏感性次之，管理类风险预警指标敏感性较低。研究结果可为相关企业建立与优化事故风险预警指标体系提供理论依据。

Screening and analysis on risk early warning indexes of production safety accidents based on graph theory

The risk early warning of accidents in the industrial production activities can effectively reduce the production accidents,but the accuracy of accident risk early warning is very dependent on the selection of accident risk early warning indexes.Taking the clinker firing workshop of a cement production enterprise as the research object,the accident causal chain theory analysis was carried out on the existing production accident cases,then a structure diagram of accident causes was constructed,and 24 accident cause nodes were obtained.19 accident cause nodes were screened out as the accident risk early warning indexes by using the analysis method of node attributes in the graph theory,and the sensitivity of accident risk early warning indexes was discussed from the local and overall perspectives by applying the improved complex network concepts of shortest path betweenness and closeness to the research objects.The results showed that the sensitivity of the risk early warning indexes contained in the risk categories of human and things were high and similar,followed by the risk early warning indexes of environment category,and the sensitivity of the risk early warning indexes in the management category was lower.The research method not only provides a theoretical basis for the cement production enterprises to establish and optimize the accident risk early warning index system,but also provides a reference idea for other industries.