Targeted intervention method for unsafe behavior based on pan-scene data of subway construction

Objectives. To effectively control unsafe behavior, improve the intervention effect and excavate the value of work safety data on subway construction sites, this study proposed a targeted intervention method for unsafe behavior based on pan-scene data of subway construction. Methods. Taking subway construction workers' unsafe behavior as the intervention target, the scene data description and structure transformation were realized, and unsafe behavior was targeted through three steps: data analysis, node positioning and strategy implementation. Firstly, based on the pan-scene data theory, 393 pan-scene data samples were sorted out by representing unsafe behavior in six dimensions: time, individual behavior, unsafe action, behavior attribute, behavior trace and professional category. Secondly, the distribution characteristics and interaction rules of unsafe behavior dimensions were explored using the association rule data mining method, and the association rules were derived to locate the intervention nodes. Furthermore, 16 specific targeted intervention measures were proposed. Results. After evaluating the intervention effect, it was found that intervention items such as system construction, safety culture and safety management were all improved to varying degrees. Conclusion. The results showed that a targeted intervention method can provide new ideas for unsafe behavior modification, and realize accurate identification, intervention and management of unsafe behavior.