Research On Data Processing Of Gas Pixel Detector

X-ray detection is an important part of the era of multi-messenger astronomy.This paper mainly studies the detection of spatial X-ray polarization information,which can be used to study the physical laws in extreme gravity environment and extreme magnetic field environment.A gas pixel detector is used to detectspatial X-ray polarization.The polarization information of X-ray is reflected through counting the emission direction of photoelectron generated by the photoelectric effect between the X-ray and the working gas in the detector.At present,the projects of detecting X-ray polarization with gas pixel detector are as follows: Polar Light project,Imaging X-ray Polarimetry Explorer（IXPE）,Enhanced X-ray Timing and Polarimetry Mission（e XTP）,etc.which will produce a large amount of scientific data about gas pixel detectors.This paper mainly explores two data processing methods of gas pixel detector written by Python,namely moment analysis method and deep learning method.The first chapter mainly introduces the development history and latest achievements of space X-ray polarization detection.Four modern scientific projects using gas pixel detectors to detect space X-ray polarization are introduced,and four cases are used to analyze the physical significance of detecting space X-ray polarization.The composition of gas pixel detector and the physical basis of detecting X-ray polarization by gas pixel detector are introduced in detail.The second chapter introduces the data of gas pixel detector.The composition and data structure of scientific data of gas pixel detector are introduced from the perspectives of experiment and simulation,respectively.The third chapter introduces how to use moment analysis method to process the data of gas pixel detector.From the perspectives of mathematical principle,Python code,method optimization and result analysis,this paper introduces in detail how to write Python program and use moment analysis method to analyze the polarization of X-ray.According to the difference between experimental data and simulation data,the concrete operations of data processing using moment analysis method are introduced respectively,including data reading,data slicing,decoding and assignment,track reduction,data noise reduction,moment calculation,reconstruction of transmission direction and drawing modulation curve,etc.Finally,through four specific cases,the data processing results obtained by moment analysis method are analyzed,the advantages and disadvantages of moment analysis method are discussed.The fourth chapter introduces how to use the deep learning method to process the data of gas pixel detector.From the perspectives of the development of deep learning,the composition of convolutional neural networks,the mainstream deep learning framework,the construction of convolutional neural networks and using models for regression prediction using models,this paper introduces in detail how to use Tensor Flow framework to build convolutional neural network to process the data of gas pixel detector.According to the difference between experimental data and simulation data,this paper introduces the steps of data reading,data assignment for forming matrix,building convolutional neural network,machine learning,saving model,and using trained model for data processing,meanwhile,introduces the implementation process and Python code required for the above operations.In this paper,two data processing methods for gas pixel detector,moment analysis method and deep learning method are introduced,which can effectively analyze the polarization of measured X-ray according to the data of the gas pixel detector.According to the analysis results of the two data processing methods,it can be found that the moment analysis method has the advantages of high speed and strong stability,but it also has the disadvantage of large average error.The deep learning method has the advantages of high accuracy and good growth,but it also has the disadvantages of slow speed and poor mobility.