发表机构
Beijing ENN Fusion Energy Science and Technology Co., Ltd.; Xi’an University of Science and Technology; Xi’an Jiaotong University; Nankai University（北京新奥聚变能源科技有限公司; 西安科技大学; 西安交通大学; 南开大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究开发AI代理框架，在含11万样本的GS数据库上测试5种神经网络架构，在EXL-50U托卡马克验证，得出CNN等模型的性能，为实时等离子体控制提供基准与选型指导。
AI 中文摘要
快速可靠的等离子体平衡预测对托卡马克的实时运行与控制至关重要，但传统Grad-Shafranov（GS）求解器往往计算成本过高，难以用于实时部署。我们开发了一种AI代理框架，并在包含100000个同分布（IID）样本和10000个分布外（OOD）样本的数值GS数据库上，对五种架构（多层感知器MLP、卷积神经网络CNN、傅里叶神经算子FNO、Transformer、柯尔莫哥洛夫-阿诺德网络KAN）进行基准测试。在统一协议下，我们评估了各模型的精度、推理效率、模型缩放性与鲁棒性。我们还通过关联数值GS解、代理预测结果与标准形状编辑器参考值，在EXL-50U托卡马克上建立了装置级验证，以评估仿真与装置的一致性。这些代理模型相对于GS解的误差为10⁻³至10⁻²，而GS到装置的偏差保持在10⁻³。Transformer在IID精度上表现最佳，而CNN在精度、鲁棒性与速度之间取得了最佳平衡，TensorRT推理延迟达0.7毫秒。在未见过的等离子体几何结构与参数 regime 下，CNN与FNO展现出最强的外推稳定性，相对L₂误差为4%至5%，而归纳偏置较弱的模型则出现显著性能退化。数据与模型容量的缩放提升了插值性能，但未必能改善外推性能，揭示了模型容量与OOD泛化能力之间的权衡。总体而言，本研究为基于AI的GS预测提供了系统且装置一致的基准，并为实时等离子体控制与聚变应用中可靠代理模型的选择提供了实践指导。
英文摘要
Fast and reliable plasma equilibrium prediction is essential for real-time tokamak operation and control, but conventional Grad-Shafranov (GS) solvers are often too costly for real-time deployment. We develop an AI surrogate framework and benchmark five architectures (MLP, CNN, FNO, Transformer, and KAN) on a numerical GS database with 100,000 IID and 10,000 OOD samples. Under a unified protocol, we evaluate accuracy, inference efficiency, model scaling, and robustness. We also establish device-level validation on the EXL-50U tokamak by linking numerical GS solutions, surrogate predictions, and the standard Shape Editor reference to assess simulation-to-device consistency. The surrogates achieve errors of $10^{-3}$-$10^{-2}$ relative to GS solutions, while the GS-to-device discrepancy remains at $10^{-3}$. Transformer gives the best IID accuracy, whereas CNN offers the best balance of accuracy, robustness, and speed, reaching 0.7 ms TensorRT latency. On unseen plasma geometries and parameter regimes, CNN and FNO show the strongest extrapolation stability, with 4%-5% relative $L_2$ error, while models with weaker inductive biases degrade more substantially. Scaling data and model capacity improves interpolation but not necessarily extrapolation, revealing a trade-off between capacity and OOD generalization. Overall, this work provides a systematic, device-consistent benchmark for AI-based GS prediction and practical guidance for selecting reliable surrogates for real-time plasma control and fusion applications.
