.png){kind=icon}
.png){kind=icon}
Return
A multi-objective optimization framework for urban flood mitigation using machine learning and optimization algorithms
.png){kind=icon}
PRE
AI
DOI：10.1016/j.jenvman.2025.128147.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The optimal design of urban flood mitigation schemes is crucial for maximizing investment effectiveness under the dual pressures of climate change and urbanization. Existing research has predominantly coupled onedimensional (1D) hydrodynamic models with optimization algorithms. However, 1D models cannot adequately represent the complex dynamics of overland flow, a critical component of urban flooding. While 1D-2D coupled hydrodynamic models offer superior fidelity in simulating these processes, their prohibitive computational cost precludes their direct integration with optimization algorithms, which demand thousands of model evaluations to converge on a solution. To bridge this critical gap, this study introduces a multi-objective optimization framework that leverages a machine learning model as a computationally efficient surrogate for the 1D-2D coupled hydrodynamic model. The proposed framework was validated through a case study, and the results are compelling: (1) The machine learning surrogate model exhibits a high degree of agreement with the high-fidelity hydrodynamic model in predicting inundation maps, while achieving a computational efficiency improvement of approximately three orders of magnitude compared to the latter. (2) Compared to the original planning scheme, the final optimized solution achieved a remarkable life-cycle cost saving of & YEN;113.5 million while simultaneously enhancing flood protection. This work presents a computationally tractable pathway for making in urban flood mitigation scheme.
Keywords:
Urban flood mitigation scheme
Multi-objective optimization
Machine learning
Metaheuristic algorithm
Journal
IF:
8.4
Papers:
2.8W
Citations:
13.7W