The Construction Of Stellar Spectral Template Library And Its Application In The Measurement Of Stellar Atmospheric Parameters

The LAMOST project is an important milestone in the history of astronomy in our country.Every year,LAMOST's sky survey program produces massive celestial spectrum data.These astronomical spectrum data contain important astrophysics information.Through measurement and research,we can have a deep understanding of celestial bodies.Through the study of stellar atmospheric physical parameters,valuable information can be effectively mined from the spectrum,which has far-reaching significance for the study of celestial spectrum data.Based on this background and based on the LAMOST-DR7 celestial body spectral data,this thesis mainly studies the construction of a stellar parameter template library and the automatic measurement of stellar atmospheric parameters based on the stellar parameter template library.The main research work of this thesis is as follows:(1)Research on the Construction of Stellar Spectral Template Library.The physical parameters of stellar atmosphere,as important information of celestial body spectrum data,mainly include effective temperature(Teff),surface acceleration of gravity(Log g),metal abundance([Fe/H])and other parameters,which are the basis of celestial spectrum research.The construction of the stellar spectral template library is an important measure of the physical parameters of the stellar atmosphere.Based on the spectral resolution,signal-to-noise ratio,direction-of-sight velocity error and stellar atmospheric parameters,this paper constructs a stellar spectral template library by means of wavelength interpolation,flow denoising,and visual inspection.The parameters of the template library covers: effective temperature: 3500K-8500 K,step size 50 K,surface gravity: 0.30dex-5.00 dex,step size 0.20 dex,metal abundance degree:-2.50dex-0.60 dex,step length is 0.10 dex,signal-to-noise ratio: SN/R>10,line-of-sight velocity error: RV error<15km/s.(2)Research on stellar atmospheric parameter measurement algorithm based on template library.This thesis selects 20000,40000 and 60000 of the celestial body spectrum data to be measured for preprocessing respectively.Before that,PCA is used to reduce the dimensionality of the stellar parameter template library,and the preprocessed celestial body spectrum to be measured is mapped to the stellar parameter template library data.And finally use Gradient boosting and Random Forests algorithms to measure the effective temperature(Teff),surface acceleration of gravity(Log g)and metal abundance([Fe/H])of the 20000,40000 and 60000 celestial body spectral data under test respectively.The parameters are measured for stellar atmospheric parameters and compared with the support vector regression(SVR)algorithm.The experimental results show that the stellar parameter template library and Gradient boosting and Random Forests algorithms can be used for automatic measurement of stellar atmospheric parameters and the prediction effect is good.