Highway traffic flow prediction model with multi-component spatial-temporal graph convolution networks.

Area of Science:

• Operations Research
• Public Health Management
• Disaster Management

Background:

• Sudden epidemic disasters cause issues like redundant or unbalanced medical material allocation, high distribution costs, and asymmetrical distribution due to poor prediction.
• Existing allocation models struggle to address the dynamic and uncertain nature of medical material demand during emergencies.

Purpose of the Study:

• To develop a robust two-stage allocation model for medical materials during sudden epidemic disasters.
• To minimize emergency response time and total allocated materials while ensuring symmetrical distribution.
• To address the challenges of prediction inaccuracies in disaster scenarios.

Main Methods:

• Introduction of prospect theory to establish a two-stage robust allocation model for medical materials.
• Proposal of the HQDRO (Heuristic Queueing-based Dynamic Resource Optimization) method based on the two-stage decision model.
• Construction of a perception model incorporating demand prediction, symmetry optimization, targeted distribution, and psychological expectations.
• Development of a scenario-based two-stage robust planning model for medical materials.

Main Results:

• The proposed two-stage robust allocation model and HQDRO effectively address issues of redundancy, imbalance, high costs, and asymmetry in medical material distribution.
• The model successfully minimizes emergency response time and total material allocation under dynamic demand constraints.
• Comparative analysis validated the effectiveness of the robust configuration model and HQDRO against existing algorithms.

Conclusions:

• The developed two-stage robust allocation model and HQDRO offer a superior approach to medical material management during epidemic disasters.
• The model ensures more symmetrical and efficient distribution of critical resources, improving emergency response capabilities.
• Integrating prospect theory and demand prediction enhances the resilience and effectiveness of disaster resource allocation.