17–22 May 2026
C.I.D
Europe/Zurich timezone

Neural network-based amplitude feedforward compensation algorithm for LLRF systems

MOP6697
18 May 2026, 16:00
2h
C.I.D

C.I.D

Deauville, France
Board: Monday wine: WD18
Poster Presentation MC6.T27: Instrumentation: Low Level RF Poster session

Speaker

Prof. Xiaofang Hu (University of Science and Technology of China)

Description

In free-electron laser facilities, the amplitude-phase flatness of the Radio Frequency (RF) pulses driving the electron beam is a key factor determining beam energy spread. Imperfections in the RF driving chain, such as the static nonlinearities of the vector modulator and the high-power amplifier, can significantly increase the intra-pulse flattop amplitude error and degrade the pulse flatness. To address this, this paper proposes a U-Net deep neural network-based amplitude feedforward compensation algorithm for Low-Level Radio Frequency (LLRF) systems to suppress intra-pulse amplitude fluctuations. The algorithm has been validated at the output of a Solid-State Amplifier (SSA): under four randomly selected LLRF output configurations (amplitude, pulse width, and pulse delay), the average intra-pulse amplitude flatness (RMS) was reduced from 1.208 % to 0.398 %, and the average peak-to-peak variation was reduced from 4.683 % to 1.353 %, demonstrating a significant compensation effect

Paper status Resubmitted proceeding files received and assigned to an editor. Accepted by Submitter.

Authors

Yuchen Wang (University of Science and Technology of China) Prof. Xiaofang Hu (University of Science and Technology of China) Mr Shenghua Yang (University of Science and Technology of China) Jian Pang (University of Science and Technology of China) Fangfang Wu (University of Science and Technology of China) Kai Zhang (University of Science and Technology of China) Shancai Zhang (University of Science and Technology of China)

Presentation materials