基于贝叶斯网络的高处坠落事故致因分析

Cause Analysis of Falling Accident Based on Bayesian Network

摘要: 随着中国经济的迅猛发展，建筑行业内的新技术和新工艺同样迅速进步，导致施工挑战性日益提升，安全生产问题频发，并且高处坠落事故是建筑业中最高发的事故。本文以1267起建筑高处坠落事故调查报告为分析对象，基于HFACS理论和TF-IDF算法，整理归纳出导致高处坠落事故的4个层次原因、15个致因因素，构建建筑高处坠落事故致因体系。然后基于apriori算法计算关联规则，得到各致因因素间的关联关系及其强度。最后，利用关联关系进一步建立贝叶斯网络模型，并利用逆向推理、致因的敏感性分析明确引起高处坠落事故的关键致因、敏感致因和一般致因，以及导致事故的最大的因素影响路径，根据模型结果提出相关的管理措施。根据研究结果可以找到高处坠落事故的源头致因，并根据致因等级可为施工单位的安全管理工作提供有效参考，做出针对性的预防措施，以防止事故的发生。

Abstract: With the rapid development of China’s economy, the construction industry is also developing rapidly, a variety of new technologies, new processes emerge in an endless stream, the construction difficulty is increasing, safety accidents occur from time to time, and height fall accident is the highest occurrence of accidents in the construction industry. This paper takes 1267 investigation reports of height fall accidents as the analysis object, and based on HFACS theory and TF-IDF algorithm, summarizes 4 levels of causes and 15 causes of height fall accidents, and constructs an identification index system of building collapse causes. Then, the association rules are calculated based on apriori algorithm to obtain the correlation and intensity among the factors. Finally, the Bayesian network model is further established by using correlation relationship, and the key causes, sensitive causes and common causes of height fall accidents are identified by reverse reasoning and sensitivity analysis of causes, as well as the influence path of the biggest factors leading to accidents, and relevant management measures are proposed according to the results of the model. According to the results of the study, the source causes of height fall accidents can be found, and according to the cause level, it can provide effective reference for the safety management of construction units, and make targeted preventive measures to prevent accidents.