Congratulations on the Recent Publication of your Research Paper, Teddy!

Paper Title: Railway Delay Prediction with Spatial-Temporal Graph Convolutional Networks

Conference: IEEE International Conference on Intelligent Transportation Systems 2020

Figures:

Figure 1 – A generic GNN architecture implementing repeated layers of graph convolutions.

Figure 2 – The overall Spatial-Temporal Graph Convolutional Network Model architecture.

Teddy’s Description:

The British rail industry is currently experiencing a stagnation in performance, affecting a rapidly growing commuter population. The Rail Research UK Association states that the total number of reactionary delay minutes per year has increased from 600,000 minutes in 2014 to 800,000 minutes in 2017. A reactionary delay is a delay to a train which results from an incident that indirectly delays the train concerned, for example, a delay to a train caused by the departure delay of another train. This type of delay can quickly spread throughout railway networks, causing further severe disruptions.

In our paper, we explored the complex, nonlinear interactions between various spatio-temporal variables that govern the propagation of delays throughout a British railway network. To better understand the effects of these nonlinear interactions, we presented a novel, graph-based formulation of a subset of the British railway network. Using this graph-based formulation, we applied the Spatial-Temporal Graph Convolutional Network Model to predict cascading delays throughout the railway network. Our finding is that this model outperforms other statistical models which do not explicitly account for interactions on the rail network, thus showing the value of a Graph Neural Network approach in predicting delays for the British railway system.

This work is supported in part by the Institute for Sustainability, Energy, and Environment (ISEE) at the University of Illinois at Urbana-Champaign and the Zhejiang University/University of Illinois at Urbana-Champaign Institute. It was led by Principal Supervisors Simon Hu and Huy T. Tran. I would also like to express gratitude to Metis Consultants who have always supported my personal development and research aspirations.

Reference: J. S. W. Heglund, P. Taleongpong, S. Hu and H. T. Tran, “Railway Delay Prediction with Spatial-Temporal Graph Convolutional Networks,” IEEE International Conference on Intelligent Transportation Systems, 2020.