Speaker
Description
Machine Learning tools are becoming part of the optimisation and control toolset at large-scale user facility such as the light source BESSY II (operated by the Helmholtz-Zentrum Berlin). The topic of this talk is the application of model-free Deep Reinforcement Learning agents for the mitigation of orbit perturbations (variations with respect to the ideal trajectory of the particles) in the storage ring. In particular, we are interested in harmonic perturbations produced by the environment - for example the main power or some imperfectly isolated magnetic sources. We will cover the design and simulation phases as well as the challenges faced during our first tests at the machine, in particular in the context of the Bluesky-based interaction framework “Naus”.