网格聚类分析天文光谱数据

发布时间：2018-05-21 10:24

本文选题：聚类分析 + FITSio　；参考：《计算机科学》2017年S2期

【摘要】：应用互联网+融合信息技术,天文大数据研究实现了海量观测数据及次生数据的高效存储、检索、数据分析及信息挖掘。现结合我国自主知识产权的大科学工程LAMOST望远镜巡天第四期(DR4)发布的经定标后的光谱数据,运用R语言中RFITSIO软件包读写光谱专用文件FITS格式,读取LAMOST发布的恒星天文数据,结合统计学和数据挖掘方法设计了有监督的网格聚类验证方案,处理并识别光谱数据,经降维提取光谱特征,归一化连续谱,保留吸收谱线特征,再划分网格聚类波长定标中心,利用相似度量函数来描述识别观测光谱数据。
[Abstract]:By using Internet fusion information technology, astronomical big data has realized the efficient storage, retrieval, data analysis and information mining of massive observation data and secondary data. In this paper, combined with the calibrated spectral data released by LAMOST Telescope, a major scientific project with independent intellectual property rights in our country, and using the FITS format of a special document for reading and writing spectra in RFITSIO software package in R language, the stellar astronomical data published by LAMOST are read and written. Combined with statistics and data mining methods, a supervised grid clustering verification scheme is designed to process and identify spectral data, extract spectral features through dimensionality reduction, normalize continuous spectra, and retain absorption spectral line features. Then the center of wavelength calibration is divided into grid clustering and the similarity function is used to describe the spectral data.
【作者单位】：齐齐哈尔大学机电工程学院;哈尔滨工程大学理学院纤维集成光学教育部重点实验室;德州学院信息管理学院;
【基金】：国家自然科学基金项目(U1631239) 黑龙江省自然科学基金资助项目(F2015203) 黑龙江省教育厅基本科研业务专项(135109219) 齐齐哈尔大学教育科学研究项目(2016072)资助
【分类号】：P11;TP311.13

【相似文献】