Description
In railway transportation, there is a high potential of energy saving by means of optimizing train schedules, circulation plans or locomotive assignments with the objective of minimizing energy consumption. There are several approaches regarding the modelling of train energy consumption, but the deterministic approach based on Davis equation is the most common method. In its basic form, the Davis equation calculates the total resistance force required to keep the train moving at a constant speed by considering the mass and velocity of the train, and journal, flange, air resistances acting on the train. In this study, data-driven energy consumption models are proposed for each locomotive type by utilising multiple linear/polynomial regressions and predicting the resistance coefficients of Davis equation. The input data is generated using two different sources: i) The real distance and gradient data for each single track in railway infrastructure is extracted from the RINF database (European Union Agency for Railways), ii) Time schedules of trains and energy consumption of locomotives are provided by the Austrian Federal Railways ( ̈OBB). The performance of each model is compared to the actual energy consumption of each locomotive type in terms of the coefficient of determination and mean squared error accuracy metrics. The Davis equations proposed in the literature with different resistance coefficient estimators are also compared to the proposed models. Finally, the provided energy consumption models are integrated into a real-world locomotive assignment optimization model.| Period | 28 Sep 2023 → 29 Sep 2023 |
|---|---|
| Event title | ÖGOR and GOR working group “Practice of Mathematical Optimization” |
| Event type | Seminar/Workshop |
| Location | Vienna, AustriaShow on map |
| Degree of Recognition | International |