| 1. |
|
|
| 2. |
|
|
| 3. |
|
|
| 4. |
- Ferreira, Diogo R., et al.
(författare)
-
High temporal resolution of pedestal dynamics via machine learning on density diagnostics
- 2024
-
Ingår i: Plasma Physics and Controlled Fusion. - 1361-6587 .- 0741-3335. ; 66:2
-
Tidskriftsartikel (refereegranskat)abstract
- At the Joint European Torus, the reference diagnostic to measure electron density is Thomson scattering. However, this diagnostic has a low sampling rate, which makes it impractical to study the temporal dynamics of fast processes, such as edge localized modes. In this work, we use machine learning to predict the density profile based on data from another diagnostic, namely reflectometry. By learning to transform reflectometry data into Thomson scattering profiles, the model is able to generate the density profile at a much higher sampling rate than Thomson scattering, and more accurately than reflectometry alone. This enables the study of pedestal dynamics, by analyzing the time evolution of the pedestal height, width, position and gradient. We also discuss the accuracy of the model when applied on experimental campaigns that are different from the one it was trained on.
|
|
| 5. |
- Fransson, Emil, 1986, et al.
(författare)
-
A fast neural network surrogate model for the eigenvalues of QuaLiKiz
- 2023
-
Ingår i: Physics of Plasmas. - 1089-7674 .- 1070-664X. ; 30:12
-
Tidskriftsartikel (refereegranskat)abstract
- We introduce a neural network surrogate model that predicts the eigenvalues for the turbulent microinstabilities, based on the gyrokinetic eigenvalue solver in QuaLiKiz. The model quickly provides information about the dominant instability for specific plasma conditions, and in addition, the eigenvalues offer a pathway for extrapolating transport fluxes. The model is trained on a 5 × 106 data points large dataset based on experimental data from discharges at the joint European torus, where each data point represents a QuaLiKiz simulation. The most accurate model was obtained when the task was split into a classification task to decide if the imaginary part of eigenvalues were stable ( ≤ 0 ) or not, and a regression model to calculate the eigenvalues once the classifier predicted the unstable class.
|
|
| 6. |
- Gillgren, Andreas, 1995, et al.
(författare)
-
Enabling adaptive pedestals in predictive transport simulations using neural networks
- 2022
-
Ingår i: Nuclear Fusion. - : IOP Publishing. - 1741-4326 .- 0029-5515. ; 62:9
-
Tidskriftsartikel (refereegranskat)abstract
- We present PEdestal Neural Network (PENN) as a machine learning model for tokamak pedestal predictions. Here, the model is trained using the EUROfusion JET pedestal database to predict the electron pedestal temperature and density from a set of global engineering and plasma parameters. Results show that PENN makes accurate predictions on the test set of the database, with R (2) = 0.93 for the temperature, and R (2) = 0.91 for the density. To demonstrate the applicability of the model, PENN is employed in the European transport simulator (ETS) to provide boundary conditions for the core of the plasma. In a case example in the ETS with varied neutral beam injection (NBI) power, results show that the model is consistent with previous studies regarding NBI power dependency on the pedestal. Additionally, we show how an uncertainty estimation method can be used to interpret the reliability of the predictions. Future work includes further analysis of how pedestal models, such as PENN, or other advanced deep learning models, can be more efficiently implemented in integrating modeling frameworks, and also how similar models may be generalized with respect to other tokamaks and future device scenarios.
|
|
| 7. |
- Gillgren, Andreas, 1995
(författare)
-
Machine learning applications for predicting the pedestal in tokamak plasmas
- 2023
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Magnetic confinement fusion is a field of research that strives to develop an environmental friendly energy source to assist in powering our society. By confining a plasma with magnetic fields, conditions that enable nuclear fusion can be achieved. However, gaining a high efficiency has proven to be a challenging task. In the 1980s, it was discovered that steep temperature and density gradients are formed near the plasma edge when the external heating passes a certain threshold leading to an increased energy and particle confinement. The region with steep gradients at the edge is referred to as the pedestal. As of today the formation and behaviour of the pedestal is still not fully understood from a theoretical standpoint. However, the enhanced performance of plasmas with a developed pedestal is routinely exploited in current fusion experiments, and is a key element in extrapolating to future devices. The purpose of this thesis is to explore machine learning methodologies to help improve the understanding and predictive capabilities of the pedestal. Specifically, a neural network for predicting pedestal characteristics has been developed and integrated with core transport models. Additionally, another neural network has been developed to enhance the temporal resolution of the main diagnostics used to analyse the pedestal. The thesis incorporates additional machine learning applications for plasma physics that extend beyond a specific focus on the pedestal.
|
|
| 8. |
- Ham, C. J., et al.
(författare)
-
Towards understanding reactor relevant tokamak pedestals
- 2021
-
Ingår i: Nuclear Fusion. - : IOP Publishing. - 1741-4326 .- 0029-5515. ; 61:9
-
Tidskriftsartikel (refereegranskat)abstract
- The physics of the tokamak pedestal is still not fully understood, for example there is no fully predictive model for the pedestal height and width. However, the pedestal is key in determining the fusion power for a given scenario. If we can improve our understanding of reactor relevant pedestals we will improve our confidence in designing potential fusion power plants. Work has been carried out as part of a collaboration on reactor relevant pedestal physics. We report some of the results in detail here and review some of the wider work which will be reported in full elsewhere. First, we attempt to use a gyrokinetic-based calculation to eliminate the pedestal top density as a model input for Europed/EPED pedestal predictions. We assume power balance at the top of the pedestal, that is, the heat flux crossing the separatrix must be equal to the heat source at the top of the pedestal and investigate the consequences of this assumption. Unfortunately, the transport assumptions of the EPED model mean that this method does not discriminate between different pairs of density and temperature profiles for a given pressure profile. Second, we investigate the effects of non flux surface density on the bootstrap current. Third, type I ELMs will not be tolerable for a reactor relevant regime due to the damage that they are expected to cause to plasma facing components. In recent years various methods of running tokamak plasmas without large ELMs have been developed. These include small and no ELM regimes, the use of resonant magnetic perturbations and the use of vertical kicks. We discuss the quiescent H-mode here. Finally we give a summary and directions for future work.
|
|
