A Nonlinear Predictive Control Approach for Urban Drainage Networks Using Data-Driven Models and Moving Horizon Estimation

Krisztian Mark Balla*, Christian Schou, Jan Dimon Bendtsen, Carlos Ocampo-Martinez, Carsten Kallesøe

*Corresponding author

Research output: Contribution to journalJournal articleResearchpeer-review


Real-time control of urban drainage networks is a complex task given their spatial dimension, nonlinear nature, and the presence of time delays imposed by non-pressurized transport flows. Unfortunately, installation of flow sensors is economically out of reach in harsh environments such as sewers, although information about volumes and flows is essential to optimize system operation and to avoid overflows. In this article, we utilize level sensors distributed throughout the network and formulate parameter and state estimation on the open-channel flow processes. A moving horizon estimation approach is proposed where the fusion of water level measurements and flow estimation is used. The aim of the presented work is twofold: to establish a data-driven, structure-preserving modeling framework leaning upon the underlying graph of the network, and to keep the model complexity within practically achievable limits, suitable for nonlinear model predictive control, which ensures easy commissioning and can handle adverse meteorological loads. The proposed control and estimation methodology are demonstrated on a physically-based high-fidelity network as virtual reality, using real rain precipitation and wastewater flow data.
Original languageEnglish
JournalI E E E Transactions on Control Systems Technology
Number of pages13
Publication statusSubmitted - 2021


  • Receding horizon control
  • Transport delays
  • Partial Differential Equations
  • Urban drainage networks
  • data-driven modeling
  • Real Time Control
  • Moving horizon estimation
  • Predictive Control
  • structure-preserving modeling
  • graph network
  • high-fidelity model
  • large-scale systems
  • multiple shooting
  • optimization
  • System Identification
  • Periodic disturbances
  • nonlinear MPC
  • State Estimation

Fingerprint Dive into the research topics of 'A Nonlinear Predictive Control Approach for Urban Drainage Networks Using Data-Driven Models and Moving Horizon Estimation'. Together they form a unique fingerprint.

Cite this