000449634 001__ 449634
000449634 005__ 20201021175339.0
000449634 0247_ $$2I:(DE-H235)DIB-20120731$$aG:(DE-HGF)2020_ZT-I-PF-5-6$$dZT-I-PF-5-6
000449634 035__ $$aG:(DE-HGF)2020_ZT-I-PF-5-6
000449634 150__ $$aAutonomous Accelerator (AA)$$y2020-2022
000449634 371__ $$0P:(DE-H253)PIP1087213$$aEichler, Annika
000449634 450__ $$aZT-I-PF-5-6$$wd$$y2020-2022
000449634 5101_ $$0I:(DE-588b)5165524-X$$aHelmholtz Gemeinschaft Deutscher Forschungszentren$$bHGF
000449634 550__ $$0G:(DE-HGF)IVF-20140101$$aImpuls- und Vernetzungsfonds$$wt
000449634 680__ $$aModern particle accelerators provide exceptional beams for new discoveries in science. The required flexibility, number of operation modes, and better performance in simultaneously more compact and more energy-efficient accelerators demand advanced control methods. One major challenge is the start-up of such accelerators, which requires frequent manual intervention. Low repetition rates, often only one acceleration event per second, lead to slow optimization rates, thus demanding expert knowledge. Although a complete autonomous accelerator seems far from being reachable, this project takes the first steps by bringing reinforcement learning to accelerator operation. Reinforcement learning yields a policy for every initial state taking the impact of the current action on the future into account, eventually replacing the need for manual intervention. This project focuses on the longitudinal bunch profile of two accelerators located at DESY and KIT. Due to the (sub)-femtosecond requirements on electron bunch duration, nonlinear and collective effects, the control of the longitudinal bunch profile is critical, challenging, and autonomous control will be indispensable for efficient and fast optimization. The high-dimensional space of adjustable parameters is posing a challenge that can only be overcome by deep reinforcement learning algorithms. Further complexity results from the continuous action and state spaces. The choice of a proper reward function is of particular importance and will be tackled here by combining the competences of machine learning experts and accelerator physicists. Results, publications as well as code, will be published open-access.
000449634 8564_ $$uhttps://hgf.desy.de/ivf/projekte/e307669/index_ger.html
000449634 909CO $$ooai:juser.fz-juelich.de:885782$$pauthority$$pauthority:GRANT
000449634 909CO $$ooai:juser.fz-juelich.de:885782
000449634 980__ $$aG
000449634 980__ $$aAUTHORITY