Research on the Model of High-Speed Railway Station Security Resource Scheduling Based on Dynamic Passenger Flow Prediction

In high-speed railway systems, the management of security resources at high-speed train stations is crucial for ensuring passenger safety and improving service efficiency. Effective security resource management enables the quick and efficient handling of large volumes of passengers, reduces queuing times, and ensures that safety measures are strictly enforced. However, current management practices often rely on a fixed-shift system, which lacks a dynamic correlation between the number of open security lanes and real-time passenger flow. This mismatch leads to resource shortages during peak times and resource wastage during off-peak periods. To address these challenges, this study introduces the Multi-Head Attention Long Short-Term Memory Network and Model Predictive Control (MHALSTM-MPC) model to improve security resource management at high-speed railway stations. The MHALSTM component predicts passenger flow by capturing trends and patterns, while the MPC component formulates an optimization problem that minimizes waiting times and operational costs by repeatedly solving it within a finite time horizon based on predicted passenger flow. This approach ensures real-time adjustments to security checkpoint configurations and staff allocation, achieving optimal resource utilization in response to forecasted demand. Experimental results based on real passenger flow data from Z City East Station demonstrate that the MHALSTM-MPC model reduces the average waiting time per passenger by 18.79% compared to the fixed-shift model and by 13.59% compared to the static scheduling model. Additionally, it achieves a 4.82% reduction in total human-hours compared to the fixed-shift model and a 2.65% reduction compared to the static scheduling model, highlighting its effectiveness in optimizing resource allocation and improving operational efficiency.