Real-time Urban Flood Management Using an Improved Genetic Algorithm-Assisted Model Predictive Control Approach

Abstract

With the increasing severity of urban flooding and related disasters, dynamic management of urban drainage systems has become increasingly critical in addressing these challenges. Model Predictive Control (MPC), as an advanced real-time control technique, has shown considerable promise; however, its performance is highly dependent on the computational efficiency of the optimization algorithm. Traditional genetic algorithms (GA) often suffer from slow convergence and low computational efficiency, which limits the real-time application of MPC. To address these challenges, this study proposed a real-time control framework for urban drainage systems that integrates an improved genetic algorithm (IGA) with MPC to enable efficient and accurate system control. The proposed approach enhanced algorithmic convergence by introducing an adaptive crossover and mutation mechanism, while significantly improving computational efficiency via parallel computing techniques. The improved genetic algorithm is embedded within the MPC framework to generate dynamic control strategies. The effectiveness of the proposed method was validated through its application to a real-world urban drainage system. The results show that:  (1) The genetic algorithm enhanced with an adaptive mechanism achieved faster convergence and demonstrated more stable convergence toward the optimal solution in the later stages of evolution. (2) The computational time of the improved GA was reduced from 305.01 minutes to 18.34 minutes under a 32-core processor environment, representing a 93.99% improvement in efficiency. (3) The MPC approach achieved an 18.93% improvement in flooding reduction, effectively mitigating the risk of urban flooding. This study provides an effective technical solution for the intelligent management of urban drainage systems, offering significant theoretical and practical value for enhancing urban flood resilience.