Physics-Informed Neural Networks for High-Precision Grad-Shafranov Equilibrium Reconstruction

2025/07/22 by Zhou, Cuizhi, Zhu, Kaien
#FOS: Physical sciences #Plasma Physics (physics.plasm-ph)

paper · doi:10.48550/arxiv.2507.16636

Abstract

The equilibrium reconstruction of plasma is a core step in real-time diagnostic tasks in fusion research. This paper explores a multi-stage Physics-Informed Neural Networks(PINNs) approach to solve the Grad-Shafranov equation, achieving high-precision solutions with an error magnitude of O(10^-8) between the output of the second-stage neural network and the analytical solution. Our results demonstrate that the multi-stage PINNs provides a reliable tool for plasma equilibrium reconstruction.

Citations

• Evaluation and Verification of Physics-Informed Neural Models of the Grad-Shafranov Equation
• A hands-on introduction to Physics-Informed Neural Networks for solving partial differential equations with benchmark tests taken from astrophysics and plasma physics
• Grad-Shafranov equilibria via data-free physics informed neural networks
• Modelling solar coronal magnetic fields with physics-informed neural networks
• Multi-stage Neural Networks: Function Approximator of Machine Precision
• Physics-informed machine learning
• Solving the Grad-Shafranov equation using spectral elements for tokamak equilibrium with toroidal rotation
• Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Physics Informed Deep Learning (Part I): Data-driven Solutions of Nonlinear Partial Differential Equations
• Accurate Derivative Evaluation for any Grad-Shafranov Solver
• On the limited memory BFGS method for large scale optimization

Related

• Grad-Shafranov equilibria via data-free physics informed neural networks
• Physics-informed Neural Operator Learning for Nonlinear Grad-Shafranov Equation
• Evaluation and Verification of Physics-Informed Neural Models of the Grad-Shafranov Equation
• CARONTE: a Physics-Informed Extreme Learning Machine-Based Algorithm for Plasma Boundary Reconstruction in Magnetically Confined Fusion Devices
• A hands-on introduction to Physics-Informed Neural Networks for solving partial differential equations with benchmark tests taken from astrophysics and plasma physics