Speaker
Description
A challenge that industrial particle accelerators face is the high amounts of noise in sensor readings. This noise obscures essential beam diagnostic and operational data, limiting the amount of information that is relayed to machine operators and beam instrumentation engineers. Machine learning-based techniques have shown great promise in isolating noise patterns while preserving high-fidelity signals, enabling more accurate diagnostics and performance tuning. Our work focuses on the implementation of a real-time FPGA-based noise reduction autoencoder, tested on a Xilinx ZCU104 evaluation kit with the intention of being deployed on industrial particle accelerators in the near future.
Funding Agency
Department of Energy, Office of Science, Office of Accelerator R&D and Production.
