Abstract Details
| status: | file name: | submitted: | by: |
|---|---|---|---|
| approved | brennan_sherwood25.pdf | 2025-02-21 13:57:09 | Dylan Brennan |
Abstracts
Author: Dylan P Brennan
Requested Type: Consider for Invited
Submitted: 2025-02-21 13:55:37
Co-authors: Cihan Akcay, K.E.J. Olofsson, John M. Finn
Contact Info:
Brennan Fusion Research
2 Hathaway Drive
Princeton Junction, NJ 08550
United States
Abstract Text:
We present a framework for estimating the probability of locking to an error field in a
rotating tokamak plasma. This framework leverages machine learning methods trained on
data from a mode-locking model, including an error field, resistive magnetohydrodynamics
modeling of the plasma, a resistive wall, and an external vacuum region, leading to a fifth-
order ordinary differential equation (ODE) system. Tearing mode saturation by a finite
island width is also modeled. We vary three pairs of control parameters in our studies:
the momentum source plus either the error field, the tearing stability index, or the island
saturation term. The order parameters are the time-asymptotic values of the five ODE
variables. Normalization of them reduces the system to 2D in control space and facilitates
the binary classification into locked (L) or unlocked (U) states, as illustrated by Akcay et
al., [Phys. Plasmas 31, 032301 (2024)]. This classification splits the control space into three
regions: ˆL, with only L states; ˆU , with only U states; and a hysteresis (hysteretic) region
ˆH, with both L and U states, where bifurcations between the L and U states can occur.
The classification of the ODE solutions into L/U is used to estimate the locking probability,
conditional on the pair of control parameters, using a neural network. A probability of
locking is a true representation of the underlying hysteresis, representing the propensity of
the mode to lock given a perturbation to the plasma. We also explore estimating the locking
probability for a sparse dataset, using a transfer learning method based on a dense model
dataset. This technique retains the qualitative characteristics of the locking probability and
allows us to obtain good estimates of it from a small number of experimental or simulation
data points, making the model highly useful in these contexts, and offering the potential for
application to real time control.
Characterization: 6.0
Comments: