Xu et al. (2025) A multi-objective optimization framework for urban flood mitigation using machine learning and optimization algorithms

Identification

• Journal: Journal of Environmental Management
• Year: 2025
• Date: 2025-12-18
• Authors: Wenbin Xu, Zheng Fang, Qianchen Xie
• DOI: 10.1016/j.jenvman.2025.128147

Research Groups

• School of Civil Engineering, Wuhan University, Wuhan, Hubei, China
• School of Resources & Environment, Nanchang University, Nanchang, Jaingxi, China

Short Summary

This study introduces a multi-objective optimization framework that leverages a machine learning model as a computationally efficient surrogate for 1D-2D coupled hydrodynamic models. The framework enables the optimal design of urban flood mitigation schemes, achieving significant cost savings and enhanced flood protection.

Objective

• To develop a computationally tractable multi-objective optimization framework for urban flood mitigation schemes by integrating a machine learning surrogate model with a 1D-2D coupled hydrodynamic model, thereby overcoming the computational cost limitations of direct integration.

Study Configuration

• Spatial Scale: Urban areas (validated through a case study, specific location not detailed in provided text).
• Temporal Scale: Life-cycle perspective for cost optimization; specific flood event durations are not detailed in the provided text.

Methodology and Data

• Models used:
  • 1D hydrodynamic models
  • 1D-2D coupled hydrodynamic models
  • Machine learning model (as a computationally efficient surrogate)
  • Multi-objective optimization algorithms
  • Metaheuristic algorithms
• Data sources: Hydrodynamic model outputs (for training the machine learning surrogate); hydrological and topographical data (implied for hydrodynamic models).

Main Results

• The machine learning surrogate model accurately predicts inundation maps, demonstrating a computational efficiency improvement of approximately 1000 times (three orders of magnitude) compared to the 1D-2D coupled hydrodynamic model.
• The optimized flood mitigation scheme achieved a life-cycle cost saving of ¥113.5 million while simultaneously improving flood protection, compared to the original planning.

Contributions

• Introduction of a novel multi-objective optimization framework that integrates a machine learning surrogate with 1D-2D coupled hydrodynamic models, enabling computationally feasible optimization of urban flood mitigation.
• Overcoming the computational barrier of high-fidelity hydrodynamic models in optimization contexts, providing a practical tool for urban planning.

Funding

• Not specified in the provided text.

Citation

@article{Xu2025multiobjective,
  author = {Xu, Wenbin and Fang, Zheng and Xie, Qianchen},
  title = {A multi-objective optimization framework for urban flood mitigation using machine learning and optimization algorithms},
  journal = {Journal of Environmental Management},
  year = {2025},
  doi = {10.1016/j.jenvman.2025.128147},
  url = {https://doi.org/10.1016/j.jenvman.2025.128147}
}