Speaker
Description
KEK-LINAC is an electron/positron linear accelerator used as the injector for the synchrotron radiation facilities (PF ring and PF-AR) and SuperKEKB. The stable operation of experiments at these facilities requires reliable beam supply from the LINAC. We have newly introduced an analytical method based on gradient boosting decision tree (GBDT) to further enhance our beam adjustment capability. GBDT is one of machine learning methods and has been used as an exceptionally effective model for tabular data. The GBDT analysis handling hundreds of LINAC operating parameters predicted accurately beam’s charge and position in the LINAC. Furthermore, by performing SHAP analysis, we have identified key parameters for the beam adjustment and correlations between the parameters. Furthermore, it was found that a model trained by the analyses can be utilized as a surrogate model for fast simulation of beam behavior in the LINAC. The results of beam adjustment with the GBDT analyses will be shown in this presentation.
| In which format do you inted to submit your paper? | LaTeX |
|---|