Design of low-carbon planning model for vehicle path based on adaptive multi-strategy ant colony optimization...

Area of Science:

• Transportation Engineering
• Artificial Intelligence
• Environmental Science

Background:

• Increasing vehicle complexity and numbers necessitate advanced route planning.
• Traffic congestion, complex routes, and high energy consumption reduce efficiency and increase pollution.
• Achieving low-carbon transportation requires strategic route optimization.

Purpose of the Study:

• To propose a novel low-carbon vehicle path planning model.
• To address the challenges of efficiency and environmental sustainability in transportation.
• To develop a model that integrates graph neural networks and ant colony optimization.

Main Methods:

• Developed an Adaptive Cooperative Graph Neural Network (ACGNN) model.
• Utilized graph data from road networks and historical trajectories as input.
• Employed subgraph screening for data quality, GNN for node/edge optimization, and Ant Colony Optimization (ACO) for global pathfinding.

Main Results:

• ACGNN demonstrated superior path planning performance compared to Dijkstra's, RGN, and conventional GNN.
• ACO significantly outperformed Simulated Annealing (SA) and Particle Swarm Optimization (PSO) in custom dataset comparisons.
• The model achieved significant improvements in path planning outcomes.

Conclusions:

• ACGNN provides an innovative technical solution for vehicle path planning.
• The proposed method effectively enhances transportation efficiency.
• ACGNN contributes to achieving low-carbon and environmentally sustainable transportation goals.