引用本文: | 高翔,刘和光,陈志敏,等.基于卷积神经网络的卫星网络协调态势评估方法[J].国防科技大学学报,2020,42(3):56-65.[点击复制] |
GAO Xiang,LIU Heguang,CHEN Zhimin,et al.Satellite networks coordination situation assessment method based on convolution neural network[J].Journal of National University of Defense Technology,2020,42(3):56-65[点击复制] |
|
|
|
本文已被:浏览 6143次 下载 4922次 |
基于卷积神经网络的卫星网络协调态势评估方法 |
高翔1,2,3,4,刘和光1,3,陈志敏1,2,姚秀娟1,2,王春梅1,2 |
(1. 中国科学院国家空间科学中心, 北京 100190;2. 中国科学院复杂航天系统电子信息技术重点实验室, 北京 100190;3. 中国科学院微波遥感技术重点实验室, 北京 100190;4. 中国科学院大学, 北京 100049)
|
摘要: |
为充分发掘利用海量卫星网络数据,提高决策效率,加强空间频轨资源获取与储备的分析手段,尤其是对地球静止轨道资源的协调获取问题,提出基于机器学习算法的卫星网络态势评估策略。通过对卫星网络协调因素进行特征分析,选择卷积神经网络(Convolution Neural Network, CNN)为目标算法模型,并建立算法模型的训练数据集及Label规则,采用分裂信息增益度量方法对数据进行降维处理,建立CNN评估模型,并进行了验证分析。结果表明,CNN模型对卫星网络协调态势评估问题测试的正确率高达80%以上,具有较高的评估效能。随着数据量的增多,CNN评估效果逐步提升,是一种在卫星网络协调态势分析、资源储备的有效评估方法。 |
关键词: 空间频轨资源 卫星网络资料 地球静止轨道 协调态势 训练集合 数据标记 卷积神经网络 |
DOI:10.11887/j.cn.202003008 |
投稿日期:2018-12-18 |
基金项目:中国科学院空间科学战略性先导专项资助项目(Y7291A1AOS);中国科学院复杂航天系统电子信息技术重点实验室开放基金资助项目(N201701) |
|
Satellite networks coordination situation assessment method based on convolution neural network |
GAO Xiang1,2,3,4, LIU Heguang1,3, CHEN Zhimin1,2, YAO Xiujuan1,2, WANG Chunmei1,2 |
(1. National Space Science Center, Chinese Academy of Sciences, Beijing 100190, China;2. Laboratory of Electronic and Information Technology for Space Systems, Chinese Academy of Sciences, Beijing 100190, China;3. Key Laboratory of Microwave Remote Sensing, Chinese Academy of Sciences, Beijing 100190, China;4. University of Chinese Academy of Sciences, Beijing 100049, China)
|
Abstract: |
In order to fully explore the use of massive satellite network data, improve decision-making efficiency,and strengthen the analysis methods of spatial frequency and orbit resource acquisition and storage, especially for the GSO (geostationary satellite orbit) resource selection problem, a satellite network situation assessment strategy based on machine learning algorithm was proposed. By analyzing the characteristics of satellite network coordination factors, the CNN (convolutional neural network) was selected as the target algorithm model, and the training data set and label rules of the algorithm model were established. The data is reduced by the split information gain measurement method and a CNN evaluation model was established. Afterwards, a verification analysis was performed. Results show that the CNN model has a correct rate of 80% or more for the satellite network coordination situation assessment problem, and has high evaluation efficiency. Moreover, with the increase of the amount of data, the evaluation effect of CNN is gradually improved, which indicates the proposed method is an effective evaluation method for coordination situation analysis and resource reserve in satellite networks. |
Keywords: space frequency and orbit resources satellite network data geostationary satellite orbit coordination situation training set data label convolution neural network |
|
|
|
|
|