Charge reconstruction of HERD silicon charge detectors based on MLP

The High Energy Cosmic-Radiation Detection (HERD) is an experimental facility designed for the study of space astronomy and particle astrophysics. The Silicon Charge Detector (SCD), as the outermost detector of HERD, has the primary objective of precisely measuring cosmic rays ranging from hydrogen to nickel. To enhance the charge resolution of the silicon charge detector by fully utilizing multi-channel information, this study employed Support Vector Machines (SVM) and Multi-Layer Perceptron (MLP) for charge reconstruction. Given the challenge of low statistics in high-Z data, we also introduced transfer learning to improve charge reconstruction for high-Z samples. Compared to our previous results (Zhanget al., 2024), the machine learning algorithm achieved an average improvement of approximately 9.8% in charge resolution for heavy nuclei with Z = 10 to Z = 28.