Abstract:Rapid urbanization and growing urban populations worldwide present significant challenges for cities, including increased traffic congestion and air pollution. Effective strategies are needed to manage traffic volumes and reduce emissions. In practice, traditional traffic flow simulations are used to test those strategies. However, high computational intensity usually limits their applicability in investigating a magnitude of different scenarios to evaluate best policies. This paper introduces an innovative approach to assess the effects of traffic policies using Graph Neural Networks (GNN). By incorporating complex transport network structures directly into the neural network, this approach could enable rapid testing of various policies without the delays associated with traditional simulations. We provide a proof of concept that GNNs can learn and predict changes in car volume resulting from capacity reduction policies. We train a GNN model based on a training set generated with a MATSim simulation for Paris, France. We analyze the model's performance across different road types and scenarios, finding that the GNN is generally able to learn the effects on edge-based traffic volume induced by policies. The model is especially successful in predicting changes on major streets. Nevertheless, the evaluation also showed that the current model has problems in predicting impacts of spatially small policies and changes in traffic volume in regions where no policy is applied due to spillovers and/or relocation of traffic.

What problem does this paper attempt to address?

The core problem that this paper attempts to solve is: how to use Graph Neural Networks (GNN) to quickly evaluate the impact of traffic policies on the reduction of urban road capacity, in order to address the problems of high computational intensity in traditional traffic flow simulation methods and the difficulty in widely testing different policy scenarios. Specifically, the research aims to provide an innovative GNN - based method. By directly incorporating the complex transportation network structure into the neural network, it can achieve rapid prediction of the effects of traffic policies, thereby providing more efficient and timely decision - support for urban planners. ### Problem Background With the acceleration of the global urbanization process, cities are facing increasingly serious traffic congestion and air pollution problems. Effective traffic management strategies are crucial for reducing vehicle use and emissions. However, although traditional traffic flow simulation methods can evaluate policy impacts in detail, their high computational complexity limits their wide application in various scenarios. Especially in small and medium - sized cities, due to the lack of data and professional knowledge, these simulation tools are often unavailable. ### Research Objectives The objective of this research is to verify whether GNN can learn and predict the effects of traffic policies on road capacity reduction. To this end, the researchers trained a GNN model using MATSim traffic simulation data in Paris and evaluated the performance of this model under different road types and scenarios. The study found that GNN can accurately predict traffic flow changes on major streets in most cases, but has difficulties when dealing with local small - scale policies or predicting traffic flow changes in areas not directly affected by policies. ### Main Contributions 1. **Innovative Method**: Proposed a GNN - based traffic policy evaluation method that can quickly predict traffic flow changes and reduce the long - time calculations required by traditional simulation methods. 2. **Practical Application**: Through a case study of Paris, proved the feasibility and effectiveness of GNN in traffic policy evaluation. 3. **Limitations and Improvement Directions**: Pointed out the deficiencies of the current model in predicting the impacts of local small - scale policies and traffic flow changes in non - policy areas, providing improvement suggestions for future research. ### Summary This paper, by introducing GNN, provides a new and efficient solution for urban traffic policy evaluation, which helps urban planners test and select the optimal traffic management strategies more quickly, thereby better coping with the challenges brought by urbanization.

Graph Neural Network Approach to Predict the Effects of Road Capacity Reduction Policies: A Case Study for Paris, France

Attention Based Graph Covolution Networks for Intelligent Traffic Flow Analysis

AGNP: Network-wide Short-Term Probabilistic Traffic Speed Prediction and Imputation

A Graph Neural Network-based Traffic Flow Prediction System with Enhanced Accuracy and Urban Efficiency

A Graph Neural Network (GNN)-Based Approach for Real-Time Estimation of Traffic Speed in Sustainable Smart Cities

Graph Neural Networks for Modelling Traffic Participant Interaction

TrafficPro: A Road Speed Prediction Framework for Urban Signal-Controlled Road Networks

Combining a Graph Neural Network Framework and a Non-Gridded Space Representation to Predict Citywide Short-Term Crash Risk

Urban Traffic Forecasting with Integrated Travel Time and Data Availability in a Conformal Graph Neural Network Framework

Simpler is better: Multilevel Abstraction with Graph Convolutional Recurrent Neural Network Cells for Traffic Prediction

Informed along the road: roadway capacity driven graph convolution network for network-wide traffic prediction

BikeNet: Accurate Bike Demand Prediction Using Graph Neural Networks for Station Rebalancing

Learning Mobility Flows from Urban Features with Spatial Interaction Models and Neural Networks

Unboxing the graph: Neural Relational Inference for Mobility Prediction

Synthesizing Traffic Datasets using Graph Neural Networks

Graph neural network surrogate for strategic transport planning

Real-Time Traffic Speed Estimation for Smart Cities with Spatial Temporal Data: A Gated Graph Attention Network Approach

GCN-GAN: Integrating Graph Convolutional Network and Generative Adversarial Network for Traffic Flow Prediction

Improving Traffic Density Forecasting in Intelligent Transportation Systems Using Gated Graph Neural Networks

Urban traffic flow prediction algorithm based on graph convolutional neural networks

A Spatiotemporal Graph Neural Network with Graph Adaptive and Attention Mechanisms for Traffic Flow Prediction