Automatic Classification of Stellar Spectra Used Neural Network

Large field spectra survey such as LAMOST will collect enormous amounts of spectra data, and these mission will require reliable and automated methods for classifying the large number of spectra data. In this paper, we present an automatic classification system of stellar spectra used neural network, it can achieve the so called MK spectral classification scheme, which is a two parameter classification, spectral type and luminosity class. The system is composed of 2 functional units, namely stellar spectra preprocessing and the classification of spectral type and luminosity class. Experiments show that the observation stellar spectra can be classed accurately in the resolution of LAMOST spectra.