Multi-Segment Variable Speed Predictive Control Strategy for Eliminating Traffic Bottlenecks in Emergency Cases on Super-Long-Span Bridges

In this paper, we investigate traffic bottlenecks, a primary cause of congestion that significantly impacts the overall efficiency of traffic networks. To address this challenge, a multi-segment variable speed limit control strategy is proposed to mitigate moving bottlenecks, particularly those on super-long-span bridges. First, an extended macro-traffic flow model, built upon the classic MetaNet framework, is proposed as a state-space model to capture the critical characteristics of long segments, which is a key contribution of this paper. Next, a fast prediction model is developed to forecast traffic flow states in lane-drop bottlenecks with restricted passing capacity over long road segments. Then, a controller leveraging the state-compensation flow model is designed to regulate the future evolution of bottleneck density. Finally, the multi-segment variable speed predictive control (MVSPC) strategy is validated on a simulation platform integrating PYTHON and SUMO, and its performance is compared with both traditional and advanced methods. The results demonstrate that under varying traffic flow levels, particularly in high-demand scenarios, the strategy achieves significant improvements in efficiency, safety, and environmental metrics. These include a 62.44% reduction in waiting time, a 95.32% decrease in potential collisions, and reductions in emissions: 26.4% in CO2, 14.11% in CO, 26.53% in NO, and 32.90% in NOX. The proposed strategy is particularly effective for long segments, such as super-long-span bridges.