| description abstract | Mega construction projects (MCPs) have long periods, complicated environments, a multiplicity of mechanical equipment and personnel, and a large number of safety risk factors on sites. The different safety risk factors of MCPs have coupling effects among them, which can produce new risks and even lead to an accident. These factors present a major challenge for project managers to take on risk management in MCPs. Although previous research has done much work on risk factor identification and control in MCPs, less attention has been given to the coupling effects among risk factors, which affect the identification effect of key safety risk factors and their coupling paths. To compensate for this gap, this research used the latent Dirichlet allocation (LDA) topic model to identify risk factors in mass accident reports, the N-K model to calculate the coupled style, and then coupled the N-K model and the social network analysis (SNA) model to identify the safety risk factors and determine the coupling paths in MCPs. The results reveal the following: (1) six first-level subsystems are identified, and there are 31 second-level safety risk factors included in MCPs; (2) H1 (low safety awareness), H2 (CAT operation), H6 (improper safety protection measures), P4 (improper safety supervisors), P5 (imperfect management system), S3 (owner’s mistakes), and M2 (improper maintenance of equipment) are the key safety risk factors in MCPs; and (3) unfavorable environmental factors can easily lead to the emergence of the multirisk coupling of G-S-P-H-M-E, which should be carried out by special management. This research provides a reference for the identification and control of safety risks in MCPs and contributes to the development of complex system risk management and control research. | |
