Multi-scenario Model Predictive Control based on genetic algorithms for level regulation of open water systems under ensemble forecasts

Xin Tian, Yuxue Guo, Rudy R. Negenborn, Lingna Wei*, Nay Myo Lin, José María Maestre

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

16 Citations (Scopus)
48 Downloads (Pure)

Abstract

Operational water resources management needs to adopt operational strategies to re-allocate water resources by manipulating hydraulic structures. Model Predictive Control (MPC) has been shown to be a promising technique in this context. However, we still need to advance MPC in the face of hydrological uncertainties. This study makes the first attempt to combine Multi-Scenario MPC (MSMPC) with a Genetic Algorithm (GA) to find Pareto optimal solutions for a multi-scenario operational water resources management problem. Then three performance metrics are adopted to select the solution to be implemented. In order to assess the performance of the proposed approach, a case study of the North Sea Canal in the Netherlands is carried out, in which ensemble discharge forecasts are used. Compared with classic MSMPC approaches that deal with uncertainty by the weighted sum approach, GA-MSMPC can better fulfill management goals although it may also be computationally expensive. With the rapid development of multi-objective evolutionary algorithms, our study suggests the potential of GA-MSMPC to deal with a wide range of operational water management problems in the future.
Original languageEnglish
Pages (from-to)3025-3040
JournalWater Resources Management
Volume33
Issue number9
DOIs
Publication statusPublished - 2019

Bibliographical note

Accepted Author Manuscript

Keywords

  • Ensemble forecasts
  • Genetic algorithms
  • Model predictive control
  • Multiple scenarios
  • Water level regulation

Fingerprint

Dive into the research topics of 'Multi-scenario Model Predictive Control based on genetic algorithms for level regulation of open water systems under ensemble forecasts'. Together they form a unique fingerprint.

Cite this