Carbon-Efficient Scheduling in Fresh Food Supply Chains with a Time-Window-Constrained Deep Reinforcement Learning...

Area of Science:

• Intelligent Transportation Systems (ITSs)
• Supply Chain Management
• Operations Research

Background:

• Intelligent Transportation Systems (ITSs) integrate Internet of Things (IoT) for enhanced vehicle, infrastructure, and user connectivity, optimizing traffic flow.
• Fresh food supply chains face challenges in distribution, including time sensitivity, temperature control, and environmental impact.
• Minimizing logistics costs and carbon emissions is crucial for sustainable fresh food distribution.

Purpose of the Study:

• To develop an optimization model for fresh food supply chain distribution routes.
• To minimize total distribution costs, including carbon emission costs and cooling expenses.
• To enhance decision-making for optimizing transportation and distribution of fresh products.

Main Methods:

• Constructed an optimization model incorporating carbon taxes for emission costs, time windows, and cooling expenses.
• Utilized a graph attention network to represent node locations, paths, and data collection windows for path planning.
• Integrated a time-window-constrained reinforcement learning model to solve for optimal distribution routes.

Main Results:

• The proposed model effectively optimizes distribution routes for fresh food products.
• Demonstrated significant reductions in logistics costs and carbon emissions.
• Provided effective decision-making information for supply chain management under varying temperature conditions.

Conclusions:

• The time-window-constrained reinforcement learning model offers an effective solution for optimizing fresh food distribution.
• The model successfully balances cost reduction, time constraints, and environmental sustainability.
• ITSs and IoT technologies are vital for improving the efficiency and environmental performance of fresh food logistics.