危化品事故人为因素关联规则挖掘与分析

Human factor association rules mining and analysis for hazardous chemical accidents

为加强生产企业危化品的安全管理,采用Apriori算法挖掘危化品事故中人为因素关联规则;以217篇危险化学品事故报告作为数据样本,基于事故统计数据,采用Apriori算法获取895条关联规则,其中,规则涵盖人为因素、事故发生时间、月份与泄漏、爆炸、火灾、中毒窒息4种事故类型的关联性;在人为因素分析结果的基础上,绘制非人为因素与违规动火作业提升度指向图,并绘制非人为因素与人为因素提升度指向图。结果表明:爆炸与火灾事故、泄漏与中毒窒息事故相互间具有强关联规则。每个事故具有相应的强关联事故时间和月份。违规动火作业行为与爆炸、火灾、泄漏3种类型事故均形成强关联规则。作业手续管理不到位、混合爆炸气体/物环境、盲板被拆除/未安装等9种非人为因素与违规动火作业产生强关联规则。

Apriori algorithm was used to mine the association rules of human factors in hazardous chemical accidents.217 accident reports were employed as data samples.Based on the accident statistics,895 association rules were obtained by Apriori algorithm.The rules cover the correlation between human factors,accident time and month and four types of accidents:leakage,explosion,fire and poisoning and asphyxiation.Based on the analysis results,a graph of non-human factors and illegal fire operation lift direction was drawn in this study.Furthermore,a graph of non-human factors and human factors lift direction was drawn.The results show that explosion and fire accidents,leakage and poisoning and asphyxiation accidents have strong correlation rules with each other.Each accident has a strong correlation with the accident time and month.Illegal fire work and three types of accidents(explosion,fire,leakage)are forming strong correlation rules.Nine types of non-human factors,such as inadequate management of work procedures,mixed explosive gas/physical environment,and removed/uninstalled blind,have strong association rules with illegal fire work.