Speaker
Description
Operational efficiency at accelerator facilities could be significantly improved through the automation of beam production and delivery, thereby increasing scientific output and nuclear data production. In this presentation, a practical path toward deploying artificial intelligence and machine learning (AI/ML) tools for the optimization of stable and radioactive ion beam production and transport at the ATLAS accelerator facility is described, with emphasis placed on the tools developed for the nuCARIBU source. The approach taken toward safe automation will be shared, including the operational safeguards by which reliable performance is supported. Lessons learned during deployment will be highlighted, and synergistic beam-tuning automation efforts at ATLAS will be discussed. How these tools and lessons could be adapted to other facilities, including university-based accelerator laboratories, will also be considered, with the aim of supporting and accelerating ongoing automation efforts across the community.
This work was supported by the Department of Energy, Office of Science, Office of Nuclear Physics, under Contract No. DE-AC02-06CH11357 and under Award No. DE-FOA-0002875. This research used resources of ANL’s ATLAS facility, which is a DOE Office of Science User Facility.