Partition independent control and collaborative optimization of high-density crowd in subway stations

The high-density and multi-directional passenger flow in subway stations has caused practical problems related to facility overload and potential safety hazards, while also posing challenges to the coordination and control methods for high-density crowd dynamics in complex systems. The dynamic evolution of dense passenger flow inside the station is a typical complex system behavior, and the streamlined design is diverse. Compared with the flow control only at the entrance of the subway, there is relatively little research on the collaborative control and optimization of flow between the entrance and the bottleneck inside the station, and there is also a problem of response lag. Given the demand for collaborative management of high-density crowd safety in complex subway systems, we propose a partition-independent control and collaborative optimization (PICCO) model based on system identification and model predictive control (MPC) to improve travel efficiency without causing new congestion. The PICCO model deduces passenger movements at the microscopic level within a complex system and models the relationship between facility control and crowd dynamics at the macroscopic level through identification methods, and the MPC method is used to predict peak passenger flow in advance and dynamically coordinate the adjustment of key area control strategies. By analyzing the predictive performance and control effectiveness of linear and nonlinear identification models and control methods under temporal response and fitness, the results show that: (i) the PICCO model effectively couples complex system modeling with predictive control for enhanced decision-making; (ii) our model emphasizes the necessity of predicting crowd density in bottleneck areas and optimizing node traffic within the station, rather than just focusing on entrance traffic control; (iii) the PICCO model excels in early identification of congestion points, dynamic adjustment of bottleneck control strategies, and multi-area collaborative control, showcasing its potential for managing complex crowd dynamics in subway stations.