3150. Implementing deep learning in digital railway control rooms
Invited abstract in session WC-39: MAI: OR in action, stream Making an Impact.
Wednesday, 12:30-14:00Room: U8
Authors (first author is the speaker)
| 1. | Marijn Verschelde
|
| Department of Economics and Quantitative Methods, IÉSEG School of Management | |
| 2. | Léon Sobrie
|
| Ghent University |
Abstract
While all predictive models rely on data, only some firms conduct predictive analytics in a full data-driven fashion via machine learning. In many business settings, predictive analytics pivot on inflexible rule-based models. However, business systems are becoming increasingly integrated, complex and heterogeneous, resulting in an exponential increase in the number of required rules. In this */presentation/*, we show the usefulness of deep learning for data-driven decision support in the context of digital railway traffic control rooms and general punctuality management; and discuss the tool we developed for the Belgian railway infrastructure company Infrabel. This tool provides visuals tailored for the traffic controller, traffic supervisor and punctuality manager at the digital control rooms. The application and near real-time implementation of our advocated DL-based predictive model are made possible by the data structure uniquely created for this project, entailing railway traffic and infrastructure data for the entire railway network.
Keywords
- Practice of OR
Status: accepted
Back to the list of papers