Search NASA⌕ Search

Engineering topics

Henderson, Morgan

Publications and source records attributed to Henderson, Morgan.

4 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Removing Noise in BPM Measurements with Variational Autoencoders

Noise in beam measurements is an ever-present challenge in accelerator operations. In addition to the challenges presented by hardware and signal processing, new operational regimes, such as ultra-short bunches, create additional difficulties in routine beam measurements. Techniques in machine learning have been successfully applied in other domains to overcome challenges inherent in noisy data. Variational autoencoders (VAEs) are shown to be capable of removing significant leevels of noise. A VAE can be used as a pre-processing tool for noise removal before the de-noised data is analyzed via other methods, or the VAE can be directly used to make beam dynamics measurements. Here we present the use of VAEs as a tool for addressing noise in BPM measurements.

43 PARTICLE ACCELERATORS↗