Position reconstruction for HERD Plastic Scintillator Detector using KAN neural network

The High Energy cosmic-Radiation Detection (HERD) facility is a dedicated high energy astronomy and particle physics experiment that will operate on the Chinese space station. Its primary objective is to detect high-energy cosmic rays (GeV $\sim$ PeV) and gamma-rays ($>\text{500}\text{MeV}$). As one of the sub-detectors of HERD, the Plastic Scintillator Detector (PSD) is designed to identify charged particles for gamma-ray detection and is also utilized to measure the charge of cosmic rays. In 2023, our first prototype of PSD was developed and tested at CERN PS and SPS beam. This paper investigates the positional response of the PSD using two reconstruction algorithms: the traditional dual-ended ratio and a deep learning approach based on KAN neural network. We find that the position resolution reconstructed by the deep learning method ranges approximately from 2.1 $\mathrm{mm}$ to 10.6 $\mathrm{mm}$ ($1\sigma$), which is significantly superior to that of the traditional method.