Author: Lundquist, J.S.
Paper Title Page
TUPAB198 ESS DTL Tuning Using Machine Learning Methods 1872
 
  • J.S. Lundquist, N. Milas, E. Nilsson
    ESS, Lund, Sweden
  • S. Werin
    Lund University, Lund, Sweden
  
  The Eu­ro­pean Spal­la­tion Source, cur­rently under con­struc­tion in Lund, Swe­den, will be the world’s most pow­er­ful neu­tron source. It is dri­ven by a pro­ton linac with a cur­rent of 62.5 mA, 2.86 ms long pulses at 14 Hz. The final sec­tion of its nor­mal-con­duct­ing front-end con­sists of a 39 m long drift tube linac (DTL) di­vided into five tanks, de­signed to ac­cel­er­ate the pro­ton beam from 3.6 MeV to 90 MeV. The high beam cur­rent and power im­pose chal­lenges to the de­sign and tun­ing of the ma­chine and the RF am­pli­tude and phase have to be set within 1% and 1 de­gree of the de­sign val­ues. The usual method used to de­fine the RF set-point is sig­na­ture match­ing, which can be a time con­sum­ing and chal­leng­ing process, and new tech­niques to meet the grow­ing com­plex­ity of ac­cel­er­a­tor fa­cil­i­ties are highly de­sir­able. In this paper we study the usage of Ma­chine Learn­ing to de­ter­mine the RF op­ti­mum am­pli­tude and phase. The data from a sim­u­lated phase scan is fed into an ar­ti­fi­cial neural net­work in order to iden­tify the needed changes to achieve the best tun­ing. Our test for the ESS DTL1 shows promis­ing re­sults, and fur­ther de­vel­op­ment of the method will be out­lined.  
DOI • reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2021-TUPAB198  
About • paper received ※ 17 May 2021       paper accepted ※ 21 June 2021       issue date ※ 13 August 2021  
Export • reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)