Algorithms for Tune Estimation and Damper Control

This Project within an internship in the radio frequency group aims to explore different methods of extracting the tune from transverse feedback and pick-up data streams. Advanced algorithms such as the extended Kalman Filter, techniques employing machine learning algorithms and the Fourier synchrosqueezed transform (FSST) are compared. Reinforcement learning is used to propose an alternative to the usual proportional feedback controller for the transverse feedback systems and performances are compared. Lastly, model predictive control (MPC) is used in combination with online identification of the accelerator's dynamics, and compared with the aforementioned proportional controller.