Computational Hydraulics
物理信息深度算子网络
Physics-Informed Deep Operator Network (DeepONet)A framework for computational hydraulics modeling that learns the solution operator of the 2D shallow water equations (SWEs) to create fast surrogate models. It operates in two modes: a purely data-driven SWE-DeepONet and a physics-informed shallow water equations DeepONet (PI-SWE-DeepONet) that incorporates continuous SWEs as constraints during training. / 一种用于计算水力建模的框架，通过学习二维浅水方程（SWE）的解算子来创建快速代理模型。它有两种运行模式：纯数据驱动的SWE-DeepONet和物理信息浅水方程DeepONet（PI-SWE-DeepONet），后者在训练过程中将连续浅水方程作为约束条件。
- Knowledge type
- 工艺技术
- First proposed or discovered
- Proposer
- Turkey Farm member
- Creator
- 农场主
- Last updated
- 2026-11-12 18:05:58
Sub-knowledge points
SWE-DeepONet: Purely data-driven learning from numerical solversPI-SWE-DeepONet: Incorporates continuous SWEs as constraints during trainingSurrogate evaluation: Less than one second on a single GPU
Higher-order knowledge points
Computational Hydraulics Modeling2D Shallow Water Equations (SWEs)Numerical Solver (e.g., SRH-2D)
Main parameters and performance indicators
inlet-discharge parameter space: normalized distance between an out-of-distribution case and the closest training sampleloss weighting approach: method for balancing competing training objectives
Open problems
Optimizing the loss weighting approachHandling the trade-off between in-distribution accuracy and out-of-distribution generalization
Related literature
ranked by community starsPhysics-Informed Deep-Operator Learning for Computational Hydraulics ModelingOpen source
Related products and links
Surrogate models for 2D hydraulic modeling
Creation and modification trail
1 recordsRecord title: Physics-Informed Deep Operator Network (DeepONet).
Node update feedback
review required