"Machine Learning for betatron tune diagnostics and control on the Super Proton Synchrotron at CERN"

"Conseil européen pour la recherche nucléaire (CERN) has made a commitment to enhance the level of automation within its accelerator facilities. The primary objectives include improving operational efficiency, increasing machine availability, and enhancing beam quality. Recent advancements in machine learning play an important role in achieving these goals. Specifically, certain critical parameters and optimization procedures, previously performed manually, have been identified as prime candidates for leveraging these technological advancements. Among these parameters are the betatron tunes and chromaticities, which are fundamental machine settings that directly impact the beam lifetime. Over the past 15 years, a tool named "Multi-Q" has been used at the Super Proton Synchrotron (SPS) to analyze and adjust these parameters. However, both the tool and the underlying analysis method have limitations in terms of accuracy, often requiring manual intervention to clean up the data. The primary objective of this project was to modernize and enhance the measurement and correction procedures. A Python package named "ml-fortune-analysis" was developed, compensating for the deficiencies of the previously employed analysis technique. Significantly improved results were achieved through a novel approach that combines HarPy spectral analysis with unsupervised machine learning. The improvements made to the tune analysis also benefited the measurement of chromaticity. The effectiveness of the new analysis method was experimentally confirmed at the SPS during beam commissioning in March 2023. The testing demonstrated improved measurement accuracy and the ability to consistently identify the tune line throughout the cycle without requiring manual intervention. Consequently, a graphical user interface (GUI) called "SPS Tune app" was developed using PyQt5 and deployed at CERN for internal use. Extensive testing of both the betatron tune and chromaticity tools was conducted at the SPS using machine development beams (MDs), yielding highly satisfactory results. The tool is now ready for use in SPS operations. The new tune analysis package has also found application in the work of other colleagues as well."