Combining Model Predictive Control with a Reduced Genetic Algorithm for Real-Time Flood Control

This paper presents a novel meta-heuristic approach for real-time flood control. The approach combines model predictive control (MPC) with a reduced genetic algorithm (RGA) to quickly find near-optimal solutions. The main control objective is to reduce the flood damage cost in an entire river basin. A fast conceptual model for the rivers and floodplains is used to compute the inundation levels and damages. The hydraulic component of the model is identified and calibrated to a full hydrodynamic model. The Demer basin, a flood-prone area in Belgium, is considered as a case study. Results show that the RGA converges faster to a near-optimal solution than a standard genetic algorithm. Furthermore, MPC-RGA outperforms the current programmable logic controller (PLC) based regulation by anticipating to rainfall forecasts.