MD3344 - Fully-Automatic Angular Alignment of LHC Collimators using Machine Learning

A complex collimation system is installed in the LHC to protect its sensitive equipment from unavoidable beam losses. An alignment procedure is used to position collimators close to the beam in the form of a hierarchy. Collimators are regularly aligned assuming no tilt between the collimator and the beam, however, tank misalignments or beam envelope angles at large-divergence locations could introduce a tilt limiting the collimation performance. Three semi-automatic angular alignment methods were introduced in an MD in 2017, however the results were not reproducible due to the number of misalignments that occurred. These methods have now been fully-automated and were successfully tested during this MD as they managed to determine the most optimal angle for each collimator without experiencing any misalignments.