EUVIMEDEUVIMED entdecken
Crossref

Automated PI Parameter Optimization for LLRF Systems: A Comparative Study of LLM-Guided Parameter Tuning and Bayesian Optimization Approaches

Lijuan Yang, Yilin Miao, Shihui Wei, Zhenglong Zhu, Zhen Gao, Zhaojie Chen, Qizhen Hou, Jiayi Peng, Jinying Ma, Chengye Xu, Xinghao Ding, Guirong Huang und 5 weitere

IEEE Transactions on Nuclear Science · 2026-05 · 73 (5) · 1830-1840

DOI: 10.1109/tns.2026.3678636 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Nuclear Science, 2026-05. Verfasst von Lijuan Yang und 16 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben. Gefördert von: Intelligent Low-Level Radio Frequency (LLRF) Control Algorithms for Superconducting Radio Frequency (RF) Cavities, National Natural Science Foundation of China, Applications of Artificial Intelligence in the Stability Study of Superconducting Linear Accelerators.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This article presents a comparative study of large language model (LLM)-guided parameter tuning and Bayesian optimization (BO) for automated PI parameter adjustment in low level radio frequency (LLRF) systems. Both methods demonstrate effective performance across continuous wave and pulsed operation modes, addressing the challenges of traditional manual tuning in superconducting accelerator facilities. Multi-cavity validation experiments were conducted on both operation modes. For continuous wave two-parameter optimization, both methods converge to optimal regions within 15 iterations. For pulsed six-parameter optimization involving coupled amplitude-phase control loops, both methods achieve satisfactory results in around 20 iterations, meeting stability requirements for beam operation. Comparative analysis reveals distinct application advantages: BO is better suited for large-scale batch optimization scenarios, particularly excelling in multi-cavity parallel commissioning and high-dimensional parameter space exploration due to its systematic probabilistic framework. The LLM-guided approach is more appropriate for online optimization scenarios with strict operational constraints, where its conservative optimization strategy and traceable reasoning process make it ideal for parameter fine-tuning during operations. The study provides practical method selection guidelines for the intelligent operation of large-scale accelerator facilities.

Förderung

  • Intelligent Low-Level Radio Frequency (LLRF) Control Algorithms for Superconducting Radio Frequency (RF) Cavities
  • National Natural Science Foundation of China
  • Applications of Artificial Intelligence in the Stability Study of Superconducting Linear Accelerators
  • Large Research Infrastructures China Initiative Accelerator Driven System Project

Zitieren

APA

Yang, L., Miao, Y., Wei, S., Zhu, Z., Gao, Z., Chen, Z., Hou, Q., Peng, J., Ma, J., Xu, C., Ding, X., Huang, G., Sun, L., Zeng, R., Wang, Z., Qiu, F., & He, Y. (2026). Automated PI Parameter Optimization for LLRF Systems: A Comparative Study of LLM-Guided Parameter Tuning and Bayesian Optimization Approaches. IEEE Transactions on Nuclear Science, 73(5), 1830-1840. https://doi.org/10.1109/tns.2026.3678636

Vancouver

Yang L, Miao Y, Wei S, Zhu Z, Gao Z, Chen Z, et al. Automated PI Parameter Optimization for LLRF Systems: A Comparative Study of LLM-Guided Parameter Tuning and Bayesian Optimization Approaches. IEEE Transactions on Nuclear Science. 2026;73(5):1830-1840. doi:10.1109/tns.2026.3678636

BibTeX
@article{Yang2026,
  author = {Yang, Lijuan and Miao, Yilin and Wei, Shihui and Zhu, Zhenglong and Gao, Zhen and Chen, Zhaojie and Hou, Qizhen and Peng, Jiayi and Ma, Jinying and Xu, Chengye and Ding, Xinghao and Huang, Guirong and Sun, Liepeng and Zeng, Rihua and Wang, Zhijun and Qiu, Feng and He, Yuan},
  title = {Automated PI Parameter Optimization for LLRF Systems: A Comparative Study of LLM-Guided Parameter Tuning and Bayesian Optimization Approaches},
  journal = {IEEE Transactions on Nuclear Science},
  year = {2026},
  volume = {73},
  number = {5},
  pages = {1830--1840},
  doi = {10.1109/tns.2026.3678636}
}
Kostenlos starten →