INFORMS STUDENT CHAPTER Seminar

Time: 10:00am-11:00am October 28

Zoom information: https://virginiatech.zoom.us/my/annosi

Speaker: Leon Sobrie, Faculty of Economics and Business Administration, Ghent University

Abstract: High-performing control room operations are paramount in a large number of industries. Control rooms act as the nerve center for real-time monitoring and intervention to coordinate many safety-critical processes in transportation, energy, healthcare and many other environments. Moreover, new technologies like AI, 5G, drones, etc. create opportunities for new types of control rooms to emerge (e.g., for smart cities and self-driving cars). Therefore, the increasing use and importance of control rooms in very heterogeneous settings imply a need for analytics tailored for control room management. Within this context, the untapped potential of machine learning for control room management is investigated in the control rooms of Infrabel, the Belgian railway infrastructure company. The highly digitized environment captures a wide variety of data on railway traffic and operator actions. We leverage the data set to analyze train delays, operator workload and human errors. Moreover, we go one step further and create predictive machine learning models for these important factors. We show that these factors can be predicted and we open the black box on the most dominant variables for predicting these factors in the studied setting. In addition, we couple the results back to the real-world setting by utilizing expert feedback to interpret the results. The developed models are implemented in real-time as a proof of concept for control room managers. With these initiatives, we aim to open a dialogue for tailored analytics in a highly digitized environment going beyond productivity by positioning the human operator at the center. We effectuate the human-centric approach by involving railway and control room experts in all process steps: problem definition, data understanding and preparation, modeling, evaluation and deployment.

Bio: Léon Sobrie is a Ph.D. Candidate at Ghent University at the Faculty of Economics and Business Administration. He holds an MSc in Business Engineering with a major in Data Analytics. He is a collaborator at the Systems Performance lab at Virginia Tech. His research focuses on developing and implementing machine learning models for predicting key metrics in digital control rooms aiming to provide real-time analytics for managers. These research efforts are executed in the context of the On Track Lab, a multi-disciplinary research lab on operations, transportation and network analytics founded by Ghent University, Infrabel and IÉSEG School of Management. His research has been presented at international conferences (INFORMS 2021 (Anaheim, USA), EURO2021 (Athens, Greece) & EURO 2022 (Espoo, Finland)).