期刊
MACHINE LEARNING-SCIENCE AND TECHNOLOGY
卷 3, 期 4, 页码 -出版社
IOP Publishing Ltd
DOI: 10.1088/2632-2153/ac93e7
关键词
collisional-radiative; plasma model; artificial neural network; adaptive sampling; active learning; surrogate model; exploration versus exploitation
类别
资金
- U.S. Department of Energy (DOE), Office of Science, Office of Advanced Scientific Computing Research [DEAC02-06CH11357]
- Office of Fusion Energy Sciences and Office of Advanced Scientific Computing Research under the Scientific Discovery through Advanced Computing (SciDAC) project of Tokamak Disruption Simulation at Los Alamos National Laboratory [89233218CNA000001]
- Office of Fusion Energy Sciences under the DeepFusion pilot project in scientific machine learning and artificial intelligence for fusion energy sciences
- US DOE Laboratory Directed Research and Development (LDRD) program [20200356ER]
- ASCR [DOE-FOA-2493]
- DOE Office of Science User Facility [DE-AC02-06CH11357]
Effective modeling of plasma transport in magnetically confined fusion devices requires accurate understanding of ion composition and radiative power losses. Artificial neural network (ANN) surrogates can be used for rapid evaluation, but training an accurate ANN relies on a large and representative dataset, which can be time-consuming to generate.
Effective plasma transport modeling of magnetically confined fusion devices relies on having an accurate understanding of the ion composition and radiative power losses of the plasma. Generally, these quantities can be obtained from solutions of a collisional-radiative (CR) model at each time step within a plasma transport simulation. However, even compact, approximate CR models can be computationally onerous to evaluate, and in-situ evaluation of these models within a larger plasma transport code can lead to a rigid bottleneck. As a way to bypass this bottleneck, we propose deploying artificial neural network (ANN) surrogates to allow rapid evaluation of the necessary plasma quantities. However, one issue with training an accurate ANN surrogate is the reliance on a sufficiently large and representative training and validation data set, which can be time-consuming to generate. In this work we explore a data-driven active learning and training routine to allow autonomous adaptive sampling of the problem parameter space to ensure a sufficiently large and meaningful set of training data is assembled for the network training. As a result, we can demonstrate approximately order-of-magnitude savings in required training data samples to produce an accurate surrogate.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据