Preparation for machine learning-aided control of HyperECR ion source
Creators
- 1. Tokyo University, Center for Nuclear Study (CNS), Tokyo (Japan)
- 2. RIKEN Nishina Center for Accelerator-Based Science, Wako, Saitama (Japan)
- 3. Rikkyo University, Graduate School of Artificial Intelligence and Science, Tokyo (Japan)
- 4. National Institute of Information and Communications Technology, Koganei, Tokyo (Japan)
- 5. RIKEN Cluster for Pioneering Research (CPR), Wako, Saitama (Japan)
Description
CNS 14 GHz Hyper ECR ion source provides various ion beams to RIKEN AVF cyclotron. It has been continuously improved for more than 30 years since its installation, and its technology for supplying high-intensity multi-charged heavy ion beams has matured. On the other hand, there is still a difficulty with beam stability, especially when supplying metal beams. Even if a sufficient current of the beam is produced stably when the ion source is tuned, the current decreases over time, and beam production becomes unstable during long-term supply. Currently, the accelerator operator adjusts the parameters against the beam fluctuation to stabilize the beam. However, it is often necessary to interrupt the experiment to adjust the beam. To solve this problem, a stability control system of the ion source is currently under development using machine learning. The preparation status of the system will be discussed in this report. (author)
Availability note (English)
Available from https://www.pasj.jp/web_publish/pasj2023/proceedings/PDF/THP3/THP37.pdfAbstract (Japanese)
東京大学CNSでは14GHz HyperECRイオン源を用いて理研AVFサイクロトロンに様々なイオンを供給している。本イオン源では設置から30年以上にわたり改良が続けられており、その多価重イオンビームの大強度供給技術は成熟してきた。一方で特に金属ビーム供給時のビーム安定度に課題が残っている。イオン源調整時に十分なビーム量が安定に出ていても、長期間の供給中にビーム量の低下やビーム生成の不安定化が起こる。現状ではビームの変動にあわせて加速器オペレータが細かくパラメータを調整することで安定化を図っているが、これには限界があり、実験を中断してビーム調整が必要になることも多い。この問題を解決するため、現在、機械学習を用いてイオン源の安定制御を補助するシステムの開発が進められている。今回の発表では、その準備状況に関して報告する。(著者)Additional details
Additional titles
- Original title (Japanese)
- 機械学習を用いたHyperECRイオン源制御の準備状況
Publishing Information
- Imprint Pagination
- [1105 p.]
- Journal Page Range
- p. 769-770
Conference
- Title
- 20. annual meeting of Particle Accelerator Society of Japan
- Acronym
- PASJ2023
- Dates
- 29 Aug - 1 Sep 2023
- Place
- Funabashi, Chiba (Japan)
INIS
- Country of Publication
- Japan
- Country of Input or Organization
- Japan
- INIS RN
- 56002046
- Subject category
- S43: PARTICLE ACCELERATORS;
- Resource subtype / Literary indicator
- Conference, Non-conventional Literature
- Descriptors DEI
- ACCELERATOR EXPERIMENTAL FACILITIES; BEAM CURRENTS; BEAM DYNAMICS; COMPUTERIZED CONTROL SYSTEMS; DATA ACQUISITION SYSTEMS; ELECTROMAGNETS; ELECTRON CYCLOTRON-RESONANCE; ION SOURCES; IPCR CYCLOTRON; MACHINE LEARNING; REGRESSION ANALYSIS; VACUUM STATES
- Descriptors DEC
- ACCELERATORS; ALGORITHMS; ARTIFICIAL INTELLIGENCE; CONTROL SYSTEMS; CURRENTS; CYCLIC ACCELERATORS; CYCLOTRON RESONANCE; CYCLOTRONS; DYNAMICS; ELECTRICAL EQUIPMENT; EQUIPMENT; HEAVY ION ACCELERATORS; ISOCHRONOUS CYCLOTRONS; LEARNING; MAGNETS; MATHEMATICAL LOGIC; MATHEMATICS; MECHANICS; ON-LINE CONTROL SYSTEMS; ON-LINE SYSTEMS; RESONANCE; STATISTICS
Optional Information
- Notes
- 3 refs., 4 figs. Imprint:#7B2C#20#56DE# #65E5##672C##52A0##901F##5668##5B66##4F1A##5E74##4F1A#