面向航天器飞控任务的大数据处理系统的设计与验证  

Design and Verification of Big Data Processing System Oriented Spacecraft Control Mission

作　　者：师明[1,2] 王保丰[1,2] 曹玉娟[1] 高宇辉[1,2] SHI Ming;WANG Baofeng;CAO Yujuan;GAO Yuhui(Beijing Aerospace Flight Control Center,Beijing 100094,China;Key Laboratory of Aerospace Flight Dynamics Technology,Beijing Aerospace Flight Control Center,Beijing 100094,China)

机构地区：[1]北京航天飞行控制中心,北京100094 [2]北京航天飞行控制中心,航天飞行动力学技术国防科技重点实验室,北京100094

出　　处：《航天器工程》2018年第6期69-76,共8页Spacecraft Engineering

摘　　要：针对传统数据组织管理方式难以满足后续大数据处理要求的问题,提出一种面向海量任务数据处理的通用型技术架构,围绕任务系统实际需求开展数据采集、数据预处理、数据存储、数据查询和数据挖掘等关键技术研究,并以开源组件、中间件和第三方组件的方式实现原型系统。分别构建单目标多测站、多目标多测站等任务场景,对系统数据采集效率、入库效率、查询能力进行用例测试验证,指标均优于传统任务系统,且满足任务使用需求;使用天宫一号数据对系统数据挖掘功能进行决策树和回归分析验证,系统目标评估指标的准确率、覆盖率都达到了100%。研究结果可为面向航天工程任务数据的组织管理、知识挖掘以及智能决策等问题提供参考。This paper puts forward a general technology architecture of big data processing,aiming at the problem that traditional data organization management is difficult to meet the requirements of follow-up big data processing,and does research on key technologies such as data acquisition,data preprocessing,data storage,data query and data mining,then implements the prototyping systems in the form of open source components,middleware,and third-party components.The task scenarios such as single-target multi-station and multi-target multi-station are constructed respectively,and the data acquisition efficiency,warehousing efficiency and query ability of the system are tested and verified by example,the indicators are superior to the traditional task system,and meet the needs of the mission requirement;the decision tree and regression analysis of the system data mining function are verified by using TG-1.data,and the accuracy and coverage of the system target evaluation index both reach 100%.The research of this paper can provide useful reference for the study of organizational management,knowledge mining and intelligent decision-making for aerospace engineering mission.

关 键 词：航天工程 大数据 分布式存储 数据挖掘 

分 类 号：V19[航空宇航科学与技术—人机与环境工程]