To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler
Müller AC, Stahlhofen P, Hammer B (2025) .
Konferenzbeitrag
| Veröffentlicht | Englisch
Download
Es wurden keine Dateien hochgeladen. Nur Publikationsnachweis!
Einrichtung
Abstract / Bemerkung
Reinforcement Learning can be a powerful tool for Pump Scheduling in Water Distribution Networks. In comparison to classic optimization it can adapt to unseen situations and find optimal schedules in real-time. In this paper, we consider the optimization of energy efficiency under pressure constraints. For this purpose, we investigate the effects of different sensory information on the learned scheduling policy. We find that information on pressure, tank levels, daytime, flows and pump energy consumption all boost the performance of the agent. However, sparse pressure readings seem to be sufficient at least in small networks.
Erscheinungsjahr
2025
Page URI
https://pub.uni-bielefeld.de/record/3016319
Zitieren
Müller AC, Stahlhofen P, Hammer B. To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler.
Müller, A. C., Stahlhofen, P., & Hammer, B. (2025). To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler. Presented at the . https://doi.org/10.18420/INF2025_151
Müller, Alissa Christin, Stahlhofen, Paul, and Hammer, Barbara. 2025. “To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler”. Presented at the . Gesellschaft für Informatik e.V.
Müller, A. C., Stahlhofen, P., and Hammer, B. (2025).“To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler”.
Müller, A.C., Stahlhofen, P., & Hammer, B., 2025. To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler.
A.C. Müller, P. Stahlhofen, and B. Hammer, “To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler”, Gesellschaft für Informatik e.V., 2025.
Müller, A.C., Stahlhofen, P., Hammer, B.: To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler. (2025).
Müller, Alissa Christin, Stahlhofen, Paul, and Hammer, Barbara. “To Pump or Not to Pump – Sensor-based Reinforcement Learning for an Optimal Scheduler”., Gesellschaft für Informatik e.V., 2025.
Google Scholar