End-to-End Heterogeneous Graph Neural Networks for Traffic Assignment

TL;DR

This paper uses the power of heterogeneous graph neural networks to propose a novel end-to-end surrogate model for traffic assignment, specifically user equilibrium traffic assignment problems, and shows that the proposed heterogeneous graph neural network model outperforms other conventional neural network models in terms of convergence rate and prediction accuracy.

Abstract

The traffic assignment problem is one of the significant components of traffic flow analysis for which various solution approaches have been proposed. However, deploying these approaches for large-scale networks poses significant challenges. In this paper, we leverage the power of heterogeneous graph neural networks to propose a novel end-to-end surrogate model for traffic assignment, specifically user equilibrium traffic assignment problems. Our model integrates an adaptive graph attention mechanism with auxiliary "virtual" links connecting origin-destination node pairs, This integration enables the model to capture spatial traffic patterns across different links, By incorporating the node-based flow conservation law into the overall loss function, the model ensures the prediction results in compliance with flow conservation principles, resulting in highly accurate predictions for both link flow and flow-capacity ratios. We present numerical experiments on urban transportation networks and show that the proposed heterogeneous graph neural network model outperforms other conventional neural network models in terms of convergence rate and prediction accuracy. Notably, by introducing two different training strategies, the proposed heterogeneous graph neural network model can also be generalized to different network topologies. This approach offers a promising solution for complex traffic flow analysis and prediction, enhancing our understanding and management of a wide range of transportation systems.

Figures (8)

• Figure 1: The illustration of the heterogeneous graph neural network for traffic assignment. The proposed model consists of three parts: graph construction & feature preprocessing module; spatial feature extraction module, and edge prediction module. The graph features are first passed into the virtual encoder (V-Encoder) through the virtual links. Then the graph embeddings are passed into the real encoder (R-Encoder) through the real links. The flow-capacity ratio and link flow of each link are calculated using the source node feature, destination node feature, and normalized edge feature.
• Figure 2: The illustrations of urban transportation networks, including Sioux Falls, EMA, and Anaheim.
• Figure 3: The histogram of the link capacity and OD demand in the training and testing data. Three transportation networks are considered including Sioux Falls, EMA, and Anaheim network.
• Figure 4: Training loss history under urban transportation network. Three benchmarks, including FCNN, GCN, and GAT, are compared with HetGAT.
• Figure 5: Illustrations of the link-wise flow distribution for different transportation networks. Three transportation networks under major disruption are considered, including Sioux Falls, EMA, and Anaheim networks. 50 links are selected for each network.