一种面向洪峰特征增强的可微分参数自适应水文建模方法
By integrating deep learning with physical hydrological models, hydrological model parameters are dynamically generated and a peak focusing loss function is introduced. This solves the problem of insufficient parameter adjustment in differentiable hydrological models under extreme rainfall conditions, and achieves high-precision simulation of peak flow and flood processes.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- DALIAN UNIV OF TECH
- Filing Date
- 2026-05-20
- Publication Date
- 2026-07-17
AI Technical Summary
Existing differentiable hydrological models are difficult to dynamically adjust parameters under extreme rainfall conditions, resulting in an underestimation of peak flow and a time lag. Furthermore, hybrid models are insufficient in simulating peak characteristics, affecting the accuracy of flood process simulation.
By fusing deep learning with physical hydrological models, hydrological model parameters are dynamically generated through recurrent neural networks, and a peak focusing loss function is introduced to construct a differentiable parameter adaptive hydrological model, enabling real-time adjustment of parameters and explicit enhancement of flood peak characteristics.
It significantly improves the simulation accuracy of peak flow and the entire flood process, solves the problem of parameter fixation in traditional models under extreme rainfall conditions, and enhances the generalization ability across watersheds.
Smart Images
Figure CN122221708B_ABST