Skip to content
Log in
#graph neural networks #diffusion models Open access

Congestion Propagation Identification and Prediction Using Self Attention–Based Diffusion Convolutional Approach

Md. Moshiur Rahman Muhammad Arif Naushin Nower
Aug 2026 · Journal of Transportation Engineering Part A Systems · 22 references
Traffic Prediction and Management Techniques Network Traffic and Congestion Control Network Security and Intrusion Detection

Abstract

Abstract Traffic congestion frequently propagates among the interconnected road networks over time, driven by spatial and temporal factors. Identifying and predicting these patterns is crucial for effective public transport management and urban planning. Existing systems often fail to accurately capture the complex diffusion properties of traffic flow and spatial dependencies, resulting in less reliable predictions. This study utilizes a propagation probability matrix to identify congestion propagation patterns and finds traffic behavior over 24 h, revealing critical insights into congestion trends in a selected road network. In addition, to overcome the remaining limitations, we propose a novel self attention–based diffusion convolutional network (SADCN) that effectively predicts traffic congestion propagation. The proposed model incorporates key spatial relations, including adjacency, diffusion, and propagation probability matrices, to improve the understanding of congestion dynamics. To demonstrate the significance of SADCN, we compare its performance with several recent graph-based models, including fully connected long short-term memory (FC-LSTM), diffusion convolutional recurrent neural network (DCRNN), and attention-based spatial-temporal graph convolutional network (ASTGCN). Compared with existing models, the proposed method achieved superior results, with an accuracy of 0.976, a precision of 0.950, a recall of 0.942, and F 1 -score of 0.946 for 50 epochs. Furthermore, the model outperformed state-of-the-art methods at shorter intervals, such as 5, 10, and 20 epochs, highlighting its faster convergence and efficiency in predicting traffic congestion propagation patterns to improve public transport systems.

DOI
View source

Similar papers

#graph neural networks Oct 2026

Locality-Preserving Graph Laplacian Manifold Learning Based Model Predictive Control for Three-Phase Inverters

This article presents a model predictive control (MPC) strategy for three-phase inverters based on locality preserving projections (LPPs). Unlike conventional machine learning–based MPC approaches that rely on predefined or high-dimensional input features, the proposed LPP-MPC automatically extracts compact, informative representations by preserving the data’s intrinsic geometric structure. This dimensionality reduction enables fast linear control-law evaluation with computational complexity O(1), making the controller well-suited for real-time implementation. Experimental results demonstrate that the LPP-MPC achieves lower total harmonic distortion (THD) and reduced tracking error compared to quadratic-programming MPC under both linear and nonlinear load conditions, and the proposed controller maintains consistently lower THD throughout load transients than other methods such as two-degree-of-freedom MPC. Compared to existing model-free MPC and deep learning neural network, the LPP-MPC has the lowest THD and root mean square error with the least computational time owing to its efficient linear structure and strong generalization capability.

Jianwu Zeng, Lizheng Cheng, V. Winstead et al. · 1 citation

Related blog posts