PaMMA-Net: Plasmas magnetic measurement evolution based on data-driven incremental accumulative prediction

2025-01-23

AI Summary

To address the poor robustness of physics-based models and the low long-horizon prediction accuracy coupled with high computational cost of existing deep learning approaches in modeling the long-term evolution of magnetic measurements for fusion plasmas, this paper proposes a deep incremental cumulative prediction framework tailored to tokamak discharge processes. We innovatively design a magnetic-measurement-specific temporal architecture and a physics-constrained data augmentation strategy, integrated with a balanced reconstruction algorithm to enhance generalization while preserving physical temporal consistency. Evaluated on real experimental data from the EAST tokamak, the method achieves over 25% improvement in prediction accuracy for magnetic signals and plasma macroscopic parameters (e.g., plasma boundary shape), reduces cross-discharge generalization error by approximately 40%, and maintains low inference overhead. This enables reliable, real-time predictive support for feedback control and underlying physics analysis in magnetic confinement fusion devices.

Technology Category

• Machine Learning: Life-Long and Continual Learning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

An accurate evolution model is crucial for effective control and in-depth study of fusion plasmas. Evolution methods based on physical models often encounter challenges such as insufficient robustness or excessive computational costs. Given the proven strong fitting capabilities of deep learning methods across various fields, including plasma research, this paper introduces a deep learning-based magnetic measurement evolution method named PaMMA-Net (Plasma Magnetic Measurements Incremental Accumulative Prediction Network). This network is capable of evolving magnetic measurements in tokamak discharge experiments over extended periods or, in conjunction with equilibrium reconstruction algorithms, evolving macroscopic parameters such as plasma shape. Leveraging a incremental prediction approach and data augmentation techniques tailored for magnetic measurements, PaMMA-Net achieves superior evolution results compared to existing studies. The tests conducted on real experimental data from EAST validate the high generalization capability of the proposed method.

Problem

• Fusion Plasma
• Model Instability
• Computation Time

Innovation

• PaMMA-Net
• Deep Learning
• Tokamak Plasma Shape Prediction