▣ Title : Machine Learning–Based Virtual Beam Diagnostics

▣ Speaker : Prof. Ji-gwang Hwang(Kangwon National University)

▣ Date & Time : 2026. 4. 2(Thu) 16:30~17:30

▣ Venue : Research Bldg I, #310

▣ Language : English

▣ Abstract :

Particle accelerators are sophisticated instruments designed to accelerate charged particles—such as electrons, protons, and heavy ions—to high energies, tailoring them for diverse applications. The precision of these experiments hinges on the ability to accurately characterize the quantitative properties of beams, including position, profile, and phase space distribution. Traditionally, beam diagnostics have relied on the analysis of the measured signal based on rigorous physical theories. However, as accelerator systems grow in complexity, theoretical modeling often requires significant approximations, leading to discrepancies in high-precision regimes. To address this, Machine Learning (ML) has emerged as a powerful alternative, enabling the interpretation of complex beam dynamics without the need for exhaustive manual modeling. This seminar presents recent advancements in Machine Learning-Based Virtual Beam Diagnostics, with a focus on the RAON (Heavy Ion Accelerator) accelerator in Korea. We demonstrate a method for predicting high-fidelity phase space distributions using minimal measurement data. Furthermore, we introduce the development of a non-invasive diagnostic system that leverages multi-button Beam Position Monitors (BPMs) integrated with ML algorithms. This approach allows for real-time, non-interceptive characterization of the beam, significantly enhancing operational efficiency and stability. Through this presentation, we will explore the synergy between accelerator physics and data science, highlighting the future of "Virtual Diagnostics" in modern high-power accelerator facilities.