This study develops and evaluates a dynamic risk-coupling assessment system for urban subway construction that identifies how human, equipment, environmental, and management risks interact over time and supports stage-specific risk control.
Risk indicators were compiled from academic and industry sources and refined through consultation with 20 subway-construction experts. The study also collected 396 questionnaire responses to evaluate the indicators and test measurement reliability and effect. Indicator weights were estimated by combining the Analytic Hierarchy Process and the Entropy Weight Method. System Dynamics was applied to simulate 24-month risk trajectories and feedback relationships, while a Cloud Model represented the fuzziness and randomness of expert judgments. Sensitivity analysis was used to examine the influence of the risk dimensions.
Human-related risk remained the dominant upstream driver throughout the simulated construction period. Equipment risk ranked second and showed a marked mid-project increase around the 12th month, creating an early-warning window for preventive maintenance and inspection. Environmental risk was comparatively stable, while management risk declined as procedures and coordination matured. The integrated model therefore supports structural decoupling of reinforcing pathways and temporal decoupling of coincident risk peaks.
The study extends existing hybrid risk-assessment approaches by linking risk-coupling theory with an interpretable System Dynamics-Cloud Model workflow. Unlike static rankings, the framework reveals feedback-driven risk trajectories, turning points, and intervention windows while explicitly representing uncertainty in expert input. It also translates these outputs into phase-specific controls for owners, designers, contractors, supervision teams, and regulators.
