The Positioning Algorithm Research For CCD Positioning Subsystem On JUNO Calibration System

The research of neutrinos plays an important role in the composition of particles and the origin of the universe.JUNO provides more precise measurements of the mass order and oscillation parameters of neutrinos by probing their energy spectrum.In order to check the performance of the PMT in JUNO,it is necessary to traverse a specified location within the detector using a calibration source,which must therefore be positioned.The active light source and the calibration source are connected together and the CCD vision positioning system obtains information on the position of the calibration source by pinpointing the active light source.Based on this the positioning algorithm research for CCD positioning subsystem on JUNO calibration system.Firstly,a combined detection system of "active light source + 4 CCD cameras" is proposed.The 4 cameras are divided into two groups,the two groups complement each other,the active light source traverses the appointed position.The 4 cameras simultaneously carry out image acquisition of the active light source,the image information collected by the 4 cameras is integrated,and the position of the active light source is accurately obtained.Secondly,in order to solve the complex environment of multi-media and multi-interface in the detection system,such as liquid scintillator,acrylic ball,pure water and air,which causes serious non-linear phenomenon.The research solution of "Zemax+Neural Network" is proposed.A simulation model is designed to obtain the correspondence between the active light source and the CCD,and a neural network structure is built to localise the position of the active light source.This solution reduces the amount of data required for CCD positioning(by a factor of approximately 1867)and increases the efficiency of CCD positioning compared to conventional CCD positioning methods,while ensuring high accuracy.Finally,a multi-media filled equivalent mini-model experimental system of acrylic,pure water and air was built to carry out experiments on the visual positioning of CCD,by comparing them with theoretical data.The experimental results show that the positioning error of the small model experimental system is less than 5 mm,verifying the effectiveness of the proposed CCD positioning system and the proposed "Zemax + neural network" research method,and meeting the positioning accuracy requirements of the JUNO calibration system.