Principal Component Analysis of Cavity Beam Position Monitor Signals

Abstract: Model-independent analysis (MIA) methods are generally useful for analysing complex systems in which relationships between the observables are non-trivial and noise is present. Principle Component Analysis (PCA) is one of MIA methods allowing to isolate components in the input data graded to their contribution to the variability of the data. In this publication we show how the PCA can be applied to digitised signals obtained from a cavity beam position monitor (CBPM) system on the example of a 3-cavity test system installed at the Accelerator Test Facility 2 (ATF2) at KEK in Japan. We demonstrate that the PCA based method can be used to extract beam position information, and matches conventional techniques in terms of performance, while requiring considerably less settings and data for calibration.