现代应用物理2025，Vol.16Issue(1)：96-104,9.DOI:10.12061/j.issn.2095-6223.202412032
朗道阻尼三阶矩方程的电子热流深度学习代理模型模拟研究
Deep Learning-Based Surrogate Model for Electron Heat Flux in Landau Damping Third-Order Moment Equations
摘要
Abstract
In recent years,closure relations for plasma fluid equations based on third-order moments have been widely studied,but direct application in fluid simulations still faces challenges.In this paper,a deep learning-based surrogate model for electron heat flux is proposed,derived from kinetic simulation data,and compared with the Hammett-Perkins closure relation.Numerical simulations show that the deep learning-based surrogate model captures the spatial distribution of electron heat flux more accurately,demonstrating higher accuracy than traditional models,especially under high wavenumber and complex disturbance conditions.When applied to third-order moment plasma fluid equations,the surrogate model effectively describes the Landau damping phenomenon under various disturbance conditions.This work provides a machine learning-based approach for coupling kinetic and fluid models,offering a new solution for multi-physics and multi-scale research in plasma.关键词
等离子体/深度学习/流体模型/代理模型/朗道阻尼/封闭条件Key words
plasma/deep learning/fluid model/surrogate model/Landau damping/closure relation分类
物理学引用本文复制引用
陈悦,张华,李明强,黄嘉昊,郑宇佳,卓红斌..朗道阻尼三阶矩方程的电子热流深度学习代理模型模拟研究[J].现代应用物理,2025,16(1):96-104,9.基金项目
国家自然科学基金资助项目(12235014,12075033) （12235014,12075033）
