MGCAF: A Novel Multigraph Cross-Attention Fusion Method for Traffic Speed Prediction.

Area of Science:

• Intelligent Transportation Systems
• Data Science
• Urban Planning

Background:

• Traffic speed prediction is crucial for efficient urban mobility and reducing vehicle emissions.
• Existing models often overlook road directionality and temporal dynamics, limiting prediction accuracy.
• Accurate traffic prediction is vital for optimizing traffic flow and mitigating environmental impact.

Purpose of the Study:

• To develop a novel model for enhanced traffic speed prediction.
• To address limitations in existing methods by incorporating road direction and temporal patterns.
• To improve the accuracy and reliability of traffic speed forecasting in urban environments.

Main Methods:

• Proposed a novel Multigraph and Cross-Attention Fusion (MGCAF) model.
• Constructed three distinct graphs: distance, positional relationships, and temporal correlations.
• Employed a multigraph attention mechanism for adaptive feature aggregation across spatial and temporal domains.

Main Results:

• The MGCAF model demonstrated superior performance in traffic speed prediction tasks.
• Experimental results on real-world datasets confirmed the model's effectiveness.
• The proposed method outperformed existing baseline approaches in prediction accuracy.

Conclusions:

• The MGCAF model effectively captures complex road network properties and temporal dependencies.
• Integrating multigraph features and cross-attention significantly enhances traffic speed prediction accuracy.
• This approach offers a promising solution for intelligent transportation systems and urban environmental management.