Developing Deep Learning Algorithms for Inferring Upstream Separatrix Density at JET

TL;DR

This paper explores machine learning methods, including supervised and semi-supervised algorithms, to predict and infer the upstream separatrix electron density in JET plasma experiments, enabling real-time core-edge plasma control.

Contribution

It introduces a novel approach combining direct mapping and representation learning for predicting $n_{e, sep}$ from JET data, including a probabilistic generative model for pedestal profiles.

Findings

Successful prediction of $n_{e, sep}$ using machine learning.

Development of a probabilistic model for pedestal profile representation.

Implementation available at GitHub.

Abstract

Predictive and real-time inference capability for the upstream separatrix electron density, $n_\text{e, sep}$, is essential for design and control of core-edge integrated plasma scenarios. In this study, both supervised and semi-supervised machine learning algorithms are explored to establish direct mapping as well as indirect compressed representation of the pedestal profiles for predictions and inference of $n_{\text{e, sep}}$. Based on the EUROfusion pedestal database for JET, a tabular dataset was created, consisting of machine parameters, fraction of ELM cycle, high resolution Thomson scattering profiles of electron density and temperature, and $n_{\text{e, sep}}$ for 608 JET shots. Using the tabular dataset, the direct mapping approach provides a mapping of machine parameters and ELM percentage to $n_{\text{e, sep}}$. Through representation learning, a compressed representation of the experimental pedestal electron density and temperature profiles is established. By conditioning the representation with machine control parameters, a probabilistic generative predictive model is established. For prediction, the machine parameters can be used to establish a conditional distribution of the compressed pedestal profiles, and the decoder that is trained as part of the algorithm can be used to decode the compressed representation back to full pedestal profiles. Although, in this work, a proof-of-principle for predicting and inferring $n_{\text{e, sep}}$ is given, such a representation learning can be used also for many other applications as the full pedestal profile is predicted. An implementation of this work can be found at https://github.com/fusionby2030/moxie.