AI-Based Control of Differential Vacuum in the MuVacAS Prototype for IFMIF-DONES Particle Accelerator
Published in EPJ Research Infrastructures, 10, 33, 2026
Recommended citation: G. Rodríguez-Llorente, G. Gallardo Romero, R. Morant Navascués, A. Sabogal, C. Torregrosa-Martín, I. Podadera, M. Weber, L. L. Bonilla, R. Gómez-Espinosa Martín. AI-Based Control of Differential Vacuum in the MuVacAS Prototype for IFMIF-DONES Particle Accelerator. EPJ Research Infrastructures, 10, 33 (2026). DOI: 10.1007/s41781-026-00186-3. https://doi.org/10.1007/s41781-026-00186-3
This open-access article presents an AI-based approach to differential-vacuum control in the MuVacAS prototype, extending the project from experimental vacuum engineering toward data-driven and intelligent control.
My contribution: engineering and experimental context from the MuVacAS vacuum-control programme.
The work combines data-driven surrogate modelling with deep reinforcement learning to estimate the inaccessible TVC pressure and regulate argon injection. Experimental validation demonstrates stable autonomous pressure control on the MuVacAS prototype, including operation with the long injection line.
