Critical Causal Path Analysis of Subway Construction Safety Accidents Based on Text Mining

Abstract

Due to the frequent occurrence of subway construction safety accidents in recent years, construction workers face great safety threat, which produces serious social hazards. The causes of subway construction safety accidents are complex and require systematic causality analysis. Therefore, this study conducts a data-driven exploration, 209 subway construction safety accident investigation reports from 2005 to 2022 are collected as the original data, and the causal paths and intrinsic mechanisms of subway construction safety accidents are analyzed, aiming to improve the safety management ability and work safety in subway construction. Firstly, a total of 116 key features associated with subway accidents are extracted from the original data by text mining, which are utilized to develop an improved Human Factors Analysis and Classification System model (HFACS). Subsequently, a novel critical causal path screening model is proposed, by combining the Apriori algorithm in the association rules algorithm with gray relational analysis. Finally, the identified critical causal paths are verified using the Jaccard similarity calculation method, which demonstrates a high degree of similarity with the original accident data and provides empirical evidence of the effectiveness and rationality of the proposed model. This study provides a new perspective for exploring the causative factors of subway construction safety accidents and the complex interaction mechanisms. Cutting off the causal propagation in the accident causal paths can eliminate potential accidents and improve the safety of subway construction. This study analyzed the causal paths and intrinsic mechanisms of subway construction safety accidents. Through text mining on 209 subway construction accident investigation reports, a total of 31 critical causes of subway construction accidents were identified. Based on the improved HFACS model, the identified causes of accidents are divided into four levels, including organizational influence, unsafe supervision, prerequisites for unsafe behavior, and unsafe behavior, in order to better analyze the main causes and their correlations of accidents. By combining the association rule algorithm and gray correlation analysis, six critical accident paths for subway construction were determined. By analyzing these accident paths, it is helpful for the management personnel to understand the complex interrelationships between critical causes and to develop targeted safety management strategies to reduce safety hazards and the likelihood of accidents, thereby improving the safety level of subway construction.