Speaker
Description
In accelerator facilities, the control and assessment of a high-quality beam delivery require capable monitoring systems, including both hardware and software components. In most accelerator beamlines, precise measurements and reliable beam delivery are critical factors in their operation. At the CERN IRRAD facility, the transverse beam profile carries the essential information about the beam properties of interest for materials and component irradiation. Precise measurements and reliable beam delivery are critical factors in its operation.
Building upon the existing IRRAD-BPM (Beam Profile Monitor) instrument at CERN, we explore the possibilities of employing Machine Learning techniques, with special focus on Autoencoder (AE) architectures. Dealing with a critical system that involves high-energy protons and extreme radiation conditions, we developed an AE-based anomaly-detection system. Its architecture, based on multiple parameters, is a result of hyperparameter optimisation aiming for the highest separation of anomalous samples. Additionally, to mitigate the existing limited BPM coverage that cannot capture the full extent of the beam tails, we perform a measurement-space statistical inference using this AE architecture. Moreover, by using a Multi-Wire Proportional Chamber (MWPC) device also present on the IRRAD beamline, we improve the beam profile modelling within a data fusion-like approach.
| In which format do you inted to submit your paper? | LaTeX |
|---|---|
| Preprint marking on your proceeding paper | I do not wish my paper to be marked as preprint. |
| I no longer wish to present this contribution, please withdraw it. | Keep my contribution |