安全与环境工程

人的不安全行为和物的不安全状态是引发建筑施工安全事故的直接原因,其中人的不安全行为占主导作用。识别工人不安全行为表现及其影响因素规律,是采取针对性安全管理措施的前提。基于文本挖掘技术和R语言,对334份建筑施工安全事故报告进行分词处理和特征项降维,识别出8项建筑工人不安全行为及其27项影响因素;通过构建共现网络和中心性分析,将27项影响因素分为关键因素、重要因素、次要因素和一般因素4个级别,其中政府安全监管职责不到位(MF₅)、安全教育培训缺失(MF₁₂)、隐患排查治理不彻底(MF₁₀)和安全检查不到位(MF₉)为4个关键因素,应对其高度重视和重点管控;最后以一起厂房坍塌事故为案例,对比分析了建筑工人不安全行为影响因素的文本挖掘分析结果与案例实际发生情况,结果发现两者具有较好的契合度,从而验证了文本挖掘和因素分级结果的实用性和合理性。

The unsafe behaviors of people and the unsafe state of objects are direct causes of construction safety accidents, among which unsafe behaviors of people play a leading role.The identification of workers' unsafe behaviors and influential factors is the basis for adopting effective safety management measures.Based on the text mining method and R language, this paper collects a total of 334 construction safety accident reports to identify eight unsafe behaviors of construction workers and 27 influential factors by using text segmentation and feature dimension reduction.Through the construction of word co-occurrence network and centrality analysis, the paper divides the 27 influential factors into four categories: critical factors, important factors, secondary factors and general factors.The result shows that there are four critical factors, including inadequate safety supervision responsibilities(MF₅),inadequate safety training(MF₁₂),delayed hazard elimination(MF₁₀),and inadequate safety inspection(MF₉),which should be attached great importance to.Finally, the paper adopts a real collapse accident in Shanghai as a case study to compare the text mining result with the actural situation of the influential factors of construction workers' unsafe behaviors.The result shows that the two are consistent, which proves the practicality of text mining result and the reasonability of factor classification result.