Beam Based Alignment Using a Neural Network

Subjects: Nuclear Science and Technology >> Radiation Physics and Technology submitted time 2024-04-07

Guanliang Wang Wei Xu

Abstract： Beams usually do not travel through the magnet centers due to errors in storage rings. The beam deviating
from the quadrupole centers is affected by additional dipole fields due to magnetic field feed-down. The beambased alignment (BBA) is often performed to find a golden orbit, on which the beam circulates around the
quadrupole center axes. For storage rings with a large number of quadrupoles, the conventional BBA procedure
is time-consuming, especially in the commissioning phase due to the necessary iterative process. Additionally,
the conventional BBA method can be affected by strong coupling and nonlinearity of the storage ring optics.
In this work, a novel method based on a neural network is proposed to find the golden orbit in a much shorter
time with reasonable accuracy. This golden orbit can be directly used for operation, or can be adopted as the
starting point for the conventional BBA. The method is demonstrated in the HLS-II storage ring for the first
time, through simulation and online experiments. The results of the experiments show that the golden orbit
obtained using this new method is consistent with that from the conventional BBA. The development of this
new method and corresponding experiments are reported in this paper.
 

YES

 Hits 591  Downloads 183  Comment 0