Transverse beam position is one of the most critical parameters in accelerator commission and operation. As non-invasive diagnostic devices, beam position monitors (BPMs) are the main “workhorse” in accelerators, providing beam center of mass position information. The position conversion factor (K-factor) of BPM systems constitutes a fundamental determinate of measurement accuracy. While precision calibration traditionally relies on moveable calibrate platforms, the prohibitive cost of equipping each BPM with a dedicated two-dimensional calibration platform remains a widespread practical constraint. In this paper, an innovative online calibration method that synergizes machine learning with beam response matrix analysis to achieve per-BPM K-factor determination is introduced. The preliminary beam experiments have been carried out at Shanghai Soft X-ray Free-Electron Laser (SXFEL) facility. The proposed method offers a robust and resource-efficient calibration solution, particularly advantageous for cavity BPM systems where conventional approaches such as theoretical calculation and offline wire scanning, fail to provide reliable results.
Chen et al. (2026) studied this question.