Using Neural Networks to Understand Field Reversed Configuration Formation Stage*

The Rotating Magnetic Field (RMF) penetration during the formation stage of a Field Reversed Configuration (FRC) can be characterized by two dimensionless parameters, $\gamma$ and $\lambda$. These two parameters account for the intensity and frequency of the RMF, as well as the plasma resistivity and number density. Work by Milroy[1], using the magnetohydrodynamics (MHD) model, has shown that in order for the RMF to penetrate a critical threshold has to be achieved. Here we use the multi-fluid plasma model and deep neural networks (DNN) to learn from simulated data the boundary between full field penetration and consequent axial field reversal and non-penetration. The results between the MHD theoretical solution and the simulated multi-fluid one are compared. Distribution A: Approved for public release; distribution unlimited; Clearance No. 18370.

 

[1] R. Milroy, “A numerical study of rotating magnetic fields as a current drive for field reversed configurations,” Phys. Plasmas, vol. 6, no. 7, pp. 2771–2780, 1999.