BriefGPT.xyz
Jan, 2024
物理约束卷积神经网络用于时空偏微分方程的逆问题
Physics-constrained convolutional neural networks for inverse problems in spatiotemporal partial differential equations
HTML
PDF
Daniel Kelshaw, Luca Magri
TL;DR
提出一种物理约束卷积神经网络(PC-CNN)来解决非线性、时空变化的偏微分方程(PDE)逆问题,演示PC-CNN在处理空间可变的拟偏误数据、重构高分辨率空间解以及分析纳维-斯托克斯方程的表现。
Abstract
We propose a
physics-constrained convolutional neural network
(PC-CNN) to solve two types of
inverse problems
in
partial differential equations
→