A Safe Bayesian Optimization Algorithm for Tuning the Optical Synchronization System at European XFEL

The European XFEL is one of the largest linear particle accelerator in the world used to generate extremely intense and ultra short X-Ray flashes to study ultra fast, time resolved chemical processes. In order to improve the quality of the observations, the laser-based optical synchronization system is optimized by tuning PI controllers using a safe Bayesian optimization approach. Since machine time on the European XFEL is very expensive, the algorithm needs to find the optimal parameters as fast as possible. In this contribution, we present a safe Bayesian optimization algorithm which, while guaranteeing safety, shows significantly improved convergence speed and noise robustness. Application and comparison results are presented in simulation for the optical synchronization system of the European XFEL and an experimental demonstration is performed for a laboratory synchronization system.

Subjects

MLE@TUHH

DDC Class

620: Engineering

Publication version

publishedVersion

Lizenz

https://creativecommons.org/licenses/by-nc-nd/4.0/

Name

1-s2.0-S2405896323018463-main-1.pdf

Size

740.16 KB

Format

Adobe PDF

Options

A Safe Bayesian Optimization Algorithm for Tuning the Optical Synchronization System at European XFEL