Speaker
Description
In the SPS, a flexible machine serving the LHC and a multitude of fixed-target experiments and fast-extraction facilities, reliable monitoring of the transverse beam position across a wide range of different beam structures and intensities is essential for stable and efficient operation. Today, the calibration procedure and signal processing of the beam position monitors (BPMs) of the SPS – and, essentially, the mapping of the digital ADC output to the analog beam signal – rely on statistical interpolation with a third-order polynomial model. This approach introduces position uncertainties reaching up to a few ~100 um. Depending on the beam intensity, structure and displacement, this compromises precision in challenging experiments, like the ones taking place in the HiRadMat facility where the beam position precision is key for the physics experiments there.
In this work, we present a machine learning-based approach for the BPMs calibration using neural networks, which aims to a more accurate mapping of the BPMs response and thus the effective reduction of these systematic errors. Using moderately deep neural networks, first results already suggest that a machine learning-based approach can outperform the previous calibration method and lead to smaller systematic errors in position measurement readings across the beam intensity and beam displacement phase space.
