EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Autonomous Water Infrastructure Management for High-Demand Facilities: A Hierarchical AI Framework Integrating Physics-Informed Digital Twins and Risk-Sensitive Reinforcement Learning

Konstantin Korvyakov Korvyakov

American Journal of Innovation in Science and Engineering · 20.11.2026 · 5 (3) · 148-153

DOI: 10.54536/ajise.v5i3.8048 ↗

Kurz gefasst

Fachartikel in der Zeitschrift American Journal of Innovation in Science and Engineering, online erschienen am 20.11.2026. Verfasst von Konstantin Korvyakov Korvyakov. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Municipal water distribution systems operating under conventional SCADA-based supervisory control lose between 28% and 35% of treated production volume before delivery to registered consumer meters, a failure rate attributable to static hydraulic model parameterization and decoupled subsystem architectures rather than to any fundamental physical constraint. This paper presents a unified framework coupling a continuously self-calibrating physics-informed digital twin, a hierarchical reinforcement learning controller decomposed via Lagrangian relaxation across three operational tiers, and a temporal fusion transformer whose probabilistic quantile outputs propagate directly into the controller’s risk-sensitive reward function. Evaluated on a 1,847-node, 312 km distribution network serving four hyperscale data center facilities over a 12-month period, the framework achieves a 33.3% reduction in non-revenue water loss, a 29.3% reduction in specific energy consumption, and a 90.7% reduction in pipe burst detection latency relative to the SCADA baseline operating on identical sensor and tariff infrastructure.

Zitieren

APA

Korvyakov, K. K. (2026). Autonomous Water Infrastructure Management for High-Demand Facilities: A Hierarchical AI Framework Integrating Physics-Informed Digital Twins and Risk-Sensitive Reinforcement Learning. American Journal of Innovation in Science and Engineering, 5(3), 148-153. https://doi.org/10.54536/ajise.v5i3.8048

Vancouver

Korvyakov KK. Autonomous Water Infrastructure Management for High-Demand Facilities: A Hierarchical AI Framework Integrating Physics-Informed Digital Twins and Risk-Sensitive Reinforcement Learning. American Journal of Innovation in Science and Engineering. 2026;5(3):148-153. doi:10.54536/ajise.v5i3.8048

BibTeX
@article{Korvyakov2026,
  author = {Korvyakov, Konstantin Korvyakov},
  title = {Autonomous Water Infrastructure Management for High-Demand Facilities: A Hierarchical AI Framework Integrating Physics-Informed Digital Twins and Risk-Sensitive Reinforcement Learning},
  journal = {American Journal of Innovation in Science and Engineering},
  year = {2026},
  volume = {5},
  number = {3},
  pages = {148--153},
  doi = {10.54536/ajise.v5i3.8048}
}
Kostenlos starten →