Iterative learning control-based adaptive beam loading compensation implementations in the J-PARC LINAC

Abstract

Abstract The Japan Proton Accelerator Research Complex (J-PARC) is a multi-purpose high-intensity proton accelerator facility that consists of a 400 MeV linear accelerator (LINAC), a 3 GeV rapid-cycling synchrotron (RCS), a 30 GeV main ring synchrotron (MR), and experimental facilities. In 2018, to achieve the goal of a 1 MW beam power at the RCS, the beam current of the LINAC was increased from 30 to 50 mA. Based on the beam loading effect, such a strong beam current can cause a significant drop in the accelerating gradients. Although both feedback and normal static feedforward control schemes were used in the low-level radio frequency (LLRF) system to suppress the beam loading effect, the peak-to-peak stability of the RF field still does not meet the requirements of the LINAC (i.e., ± 0.5% amplitude and ± 0 . 5 ° phase). To solve this problem, an iterative learning control (ILC) scheme was studied and implemented in the J-PARC LINAC. The beam loading compensation experiments demonstrate that the inclusion of the ILC controller improves the performance of the control system significantly. As the number of iterations increase, the tracking error of the system decreases monotonously. For the accelerating field with beam operation, compared to the performance with static feedforward control, the peak-to-peak stability of amplitude improves from greater than ± 1% to less than ± 0.4%, and the peak-to-peak stability of phase improves from ± 1°to ± 0 . 2 ° .