基于深度学习技术的恒星大气物理参数自动估计

发布时间：2018-06-19 00:39

  本文选题：恒星 + 基本参数　； 参考：《天文学报》2016年04期

【摘要】：深度学习是当前机器学习、模式识别和人工智能领域中的一项热点研究技术,非常适用于处理复杂的大规模数据.基于深度学习理论构建了一个5层的栈式自编码深度神经网络,对恒星大气物理参数进行自动估计,网络各层的节点数分别为3821-500-100-50-1.使用美国大型巡天项目Sloan发布的Sloan Digital Sky Survey(SDSS)实测光谱以及由Kurucz的New Opacity Distribution Function(NEWODF)模型得到的理论光谱进行了实验验证,对有效温度(Teff)、表面重力加速度(lg g)和金属丰度([Fe/H])3个物理参数进行了自动估计.结果表明,栈式自编码深度神经网络的估计精度较好,其中在SDSS数据上的平均绝对误差分别为:79.95(Teff/K),0.0058(lg(Teff/K)),0.1706(lg(g/(cm·s~(-2)))),0.1294 dex([Fe/H]);在理论数据上的平均绝对误差分别是:15.34(Teff/K),0.0011(lg(Teff/K)),0.0214(lg(g/(cm·s~(-2)))),0.0121 dex([Fe/H]).
[Abstract]:Deep learning is a hot research technology in the field of machine learning, pattern recognition and artificial intelligence, which is very suitable for dealing with complex large-scale data. Based on the depth learning theory, a five-layer self-coding depth neural network is constructed. The parameters of stellar atmosphere are estimated automatically. The number of nodes in each layer of the network is 3821-500-100-50-1. The measured spectra of Sloan Digital Sky Survey (SDSS) published by Sloan and the theoretical spectra obtained from Kurucz's New Opacity Distribution function ODF model are verified experimentally. Three physical parameters, I. e., effective temperature, surface gravity acceleration (LG) and metal abundance ([Fe / H]), are estimated automatically. 缁撴灉琛ㄦ槑,鏍堝紡鑷�缂栫爜娣卞害绁炵粡缃戠粶鐨勪及璁＄簿搴﹁緝濂�,鍏朵腑鍦⊿DSS鏁版嵁涓婄殑骞冲潎缁濆�硅��宸�鍒嗗埆涓，

本文编号：2037532