Risk Identification and Resilience Assessment of Irregular Intersections Under Cascading Failures: A Case Study of the Donggang Passenger Station Intersection

This paper proposes a resilience assessment model for irregular intersections that combines conflict analysis, channelization conditions, and traffic operational data to identify turning movements with high cascading-failure potential. Turning movements are the basic analytical unit. A channelization-weighted conflict matrix captures coupling among traffic flows by incorporating signal phase separation and lane function allocation. A CLI (Conflict Load Index) integrates the weighted conflict degree, a saturation correction factor, and expert risk scores to rank turning movements by cascading-failure risk. Three failure scenarios are compared: random failure, descending conflict-degree failure, and descending CLI failure. A failure propagation probability function based on weighted conflict-degree load distribution is paired with a resilience loss index that quantifies cumulative intersection capacity loss during failure propagation. The model is applied to the irregular intersection near Donggang Passenger Station. Among 16 turning movements, R2-T2, R4-T2, R5-T2, and R1-T1 have the four highest CLI values and form the high-risk set. The high-saturation movement R5-T2 (saturation 0.738) rises to rank 3 in the revised model, three positions higher than in the topology-only model, reflecting its risk level under actual traffic conditions. Sensitivity analysis shows that the relative ranking of the three failure modes is consistent across all parameter combinations, confirming the robustness of the conclusions. The model provides a quantitative basis for resilience diagnosis, risk early warning, and improvement planning at irregular intersections.