|
| Title: | Efficient multi-scenario model predictive control for water resources management with ensemble streamflow forecasts | | | Authors: | X. Tian, R.R. Negenborn, P.J. van Overloop, J.M. Maestre, A. Sadowska, N. van de Giesen |
| | Journal: | Advances in Water Resources | | |
| | Abstract: | Model Predictive Control (MPC) is one of the most advanced real-time control techniques that has been widely applied to water resources management (WRM). MPC can manage the water system in a holistic manner and has a flexible structure to incorporate specific elements, such as setpoints and constraints. Therefore, MPC has shown its versatile performance in many branches of WRM. Nonetheless, with the in-depth understanding of stochastic hydrology in recent studies, MPC also faces the challenge of how to cope with hydrological uncertainty in its decision-making process. A possible way to embed the uncertainty is to generate an ensemble forecast (EF) of hydrological variables, rather than a deterministic one. The combination of MPC and EF results in a more comprehensive approach: Multi-scenario MPC (MS-MPC). In this study, we will first assess the model performance of MS-MPC, considering ensemble streamflow forecasts. Noticeably, the computational inefficiency may be a critical obstacle that hinders applicability of MS-MPC. In fact, with more scenarios taken into account, the computational burden of solving an optimization problem in MS-MPC accordingly increases. To deal with this challenge, we propose the Adaptive Control Resolution (ACR) approach as a computationally efficient scheme to practically reduce the number of control variables in MS-MPC. In brief, the ACR approach uses a mixed-resolution control time step, based on the change in the forecast accuracy from the near future to the distant future. The ACR-MPC approach is tested on a real-world case study: an integrated flood control and navigation problem in the North Sea Canal of the Netherlands. Such an approach reduces the computation time by 18% and up. At the same time, the model performance of ACR-MPC remains close to that of conventional MPC. |
| | Reference: | Efficient multi-scenario model predictive control for water resources management with ensemble streamflow forecasts. X. Tian, R.R. Negenborn, P.J. van Overloop, J.M. Maestre, A. Sadowska, N. van de Giesen. Advances in Water Resources, vol. 109, pp. 58-68, 2017. | | | Request: | A
copy of this publication. |
|
|
