摘要
光谱分类识别一直是天文学家研究中的基础问题,也是LAMOST巡天计划的一项重要任务.从LAMOST发布的海量天体光谱数据库中选取F、G、K 3种型星光谱数据,采用深度学习模型进行分类识别研究和对比实验研究,解决原有方法对光谱分类可信度低的问题.实验结果证明:对于F、G、K 3种型星的分类精确度问题,深度学习方法明显优于原有其他分类方法.
Spectral classification and recognition has always been the basic problem in the astronomical studies, and it is also an important task of LAMOST sky survey project. This paper selects three types(F, G and K) of spectra data from LAMOST, and gives them a comparison of the criteria for classification and recognition based on a deep learning model in order to solve the problem of low credibility of the spectral classification by the traditional classification method. The results show that this deep learning method is superior to other classification methods for the classification accuracy of the above-mentioned types of stars.
引文
[1] 吴永东,马倾德.多尺度形态滤形及弹性匹配技术在类星体谱线识别中的应用[J].中国图象图形学报,1997,2(1):3-6.
[2] 邱波,胡占义,赵永恒.光谱描述语言和粗集[J].光谱学与光谱分析,2002,22(3):523-526.
[3] 覃冬梅,胡占义,赵永恒.一种基于主分量分析的恒星光谱快速分类法[J].光谱学与光谱分析,2003,23(1):182-186.
[4] 覃冬梅,胡占义,赵永恒.基于支撑矢量机的天体光谱自动分类方法[J].光谱学与光谱分析,2004,24(4):507-511.
[5] 罗阿理.光谱自动分析技术中的模式识别方法研究[D].北京:中国科学院国家天文台,2002.
[6] 赵瑞珍,王飞,罗阿理,等.基于稀疏表示的谱线自动提取方法[J].光谱学与光谱分析,2009,29(7):2010-2013.
[7] LUO A L,ZHAO Y H,ZHAO G,et al.The first data release (DR1) of the LAMOST regular survey[J].Research in Astronomy and Astrophysics,2015,15(8):1095-1124.
[8] LIU C,CUI W Y,ZHANG B,et al.Spectral classification of stars based on LAMOST spectra[J].Research in Astronomy and Astrophysics,2015,15(8):1137-1153.
[9] HINTON G E,SALAKHUTDINOV R R.Reducing the dimensionality of data with neural networks[J].Science,2006,313(5786):504-507.
[10] ACKLEY D H,HINTON G E,SEJNOWSKI T J.A learning algorithm for Boltzmann machines[J].Cognitive Science,1985,9(1):147-169.
[11] WANG K,GUO P,LUO A L.A new automated spectral feature extraction method and its application in spectral classification and defective spectra recovery[J].Monthly Notices of the Royal Astronomical Society,2016,465(4):4311-4324.