Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Initialising ...
Sugita, Moe; Horino, Koki*; Nomura, Masahiro; Shinozaki, Shinichi; Takayanagi, Tomohiro; Ueno, Tomoaki*; Kuriyama, Yasutoshi
Proceedings of 16th International Particle Accelerator Conference (IPAC25) (Internet), p.2715 - 2717, 2025/11
no abstracts in English
Sugita, Moe; Takayanagi, Tomohiro; Ueno, Tomoaki*; Horino, Koki*; Shinozaki, Shinichi
Proceedings of 21st Annual Meeting of Particle Accelerator Society of Japan (Internet), p.730 - 732, 2024/10
no abstracts in English
Sugita, Moe; Takayanagi, Tomohiro; Ueno, Tomoaki*; Ono, Ayato; Horino, Koki*; Kinsho, Michikazu; Oguri, Hidetomo; Yamamoto, Kazami
Proceedings of 20th Annual Meeting of Particle Accelerator Society of Japan (Internet), p.519 - 522, 2023/11
In J-PARC RCS, paint bump magnets are used to displace the beam orbit during paint injection, which produces a high intensity beam. A pattern of command current and command voltage can be used to create an output current waveform that varies the beam orbit over time. The accuracy of beam orbit control is determined by the shape difference between the command current and output current waveforms. In the current paint pattern adjustment, a deviation of 
1% or less is achieved by manual adjustment after using software that adjusts the pattern according to the response function of the power supply control. However, we would like to reduce the adjustment time. In addition, since the accuracy of paint injection is determined by the adjustment system of the paint magnet power supply, we would like to achieve output current deviation 10 times more precise than before to reduce beam loss. An analytical model of the load-side impedance is necessary to create a high-precision paint pattern, but it is very difficult to construct an analytical model because the load-side impedance changes in a time-varying nonlinear paint pattern. We used machine learning to adjust the output pattern of the paint pattern and achieved a deviation of less than 0.5% through repeated learning. This presentation will report on the current status of the system and its prospects.
Sugita, Moe; Nomura, Masahiro; Ueno, Tomoaki*; Kuriyama, Yasutoshi; Horino, Koki*; Takayanagi, Tomohiro; Shinozaki, Shinichi
no journal, ,
no abstracts in English
