Speaker
Description
High-precision DCCTs for particle accelerator beam intensity measurement rely on perfectly matched ferromagnetic core pairs to achieve excellent zero-flux stability. Even within the same batch of nanochrystalline or amorphous alloys, small variations in $H_c$, $B_r$, and $B_s$ cause unacceptable offset and noise. Conventional manual sorting by single-parameter thresholds is inefficient and suboptimal.
This paper proposes an automated core-pairing method using unsupervised machine learning. Dynamic B-H parameters from a large batch of cores were collected and clustered with K-Means, DBSCAN, and hierarchical clustering algorithms to group cores with high multidimensional similarity. Prototype modulators assembled from ML-selected pairs for Co-based amorphous alloy rings were tested, showing significantly lower zero-flux offset and noise floor than randomly or single-parameter-matched pairs. The approach provides a fast, reproducible, and scalable solution for high-performance DCCT production.
Funding Agency
- This work is supported in part by National Natural Science Fund (No. 12275294).
Footnotes
† huangwei@ihep.ac.cn
| In which format do you inted to submit your paper? | LaTeX |
|---|