SPIRAL2 Cryomodules Models: a Gateway to Process Control and Machine Learning

TL;DR

This paper demonstrates how numerical models of the SPIRAL2 cryogenic system can be used for optimal control, virtual sensing, and anomaly detection, improving system performance and reliability.

Contribution

It introduces three novel applications of cryostat modeling for control, sensing, and fault detection in particle accelerator systems.

Findings

Optimal controller synthesis improved system stability.

Virtual sensors provided accurate real-time measurements.

Anomaly detection identified faults effectively.

Abstract

From simple physical systems to full production lines, numerical models could be used to minimize downtime and to optimize performances. In this article, the system of interest is the SPIRAL2 (Syst\`eme de Production d'Ions RAdioactifs en Ligne de 2e g\'en\'eration) particles accelerator cryogenic system. This paper illustrates three totally different applications based on a SPIRAL2 cryostat model: optimal controller synthesis, virtual sensor synthesis and anomaly detection. The tow firsts applications have been deployed on the system. Experimental results are used to illustrate the benefits of such applications.