Speaker
Description
Performance drift has been a longstanding problem for accelerators. A desirable solution is to tune the machine slowly and gently to compensate for such drift. Previously, we presented a version of the Multi-Generation Gaussian Process Optimizer which tunes accelerator settings during operation to maintain optimal performance. In this paper, we present an improved version of the algorithm and its application test examples, in which it corrects deviations from the ideal orbit caused by a drifting orbit corrector magnet and a drifting injection kicker magnet respectively. The modified algorithm takes measures to ensure the accuracy of the Gaussian process regression models and to improve the validity of the new trial solutions. We demonstrate that this is a promising development toward using safe, real-time tuning algorithms during accelerator programs to compensate for performance drift.
| In which format do you inted to submit your paper? | LaTeX |
|---|