基于随机森林的航天器电信号多分类识别方法

兰巍; 贾素玲; 宋世民; 李可

doi:10.13700/j.bh.1001-5965.2016.0661

基于随机森林的航天器电信号多分类识别方法

doi: 10.13700/j.bh.1001-5965.2016.0661

兰巍¹,
贾素玲¹,
宋世民²,
李可^3, ,

1.
北京航空航天大学经济管理学院, 北京 100083
2.
中国空间技术研究院, 北京 100194
3.
北京航空航天大学航空科学与工程学院, 北京 100083

基金项目:

航空科学基金 2012XX1043

中央高校基本科研业务费专项资金 YWF-16-HKXY-017

国家自然科学基金 61773039

详细信息

作者简介:
兰巍   女, 博士研究生; 主要研究方向:数据挖掘与机器学习

贾素玲   女, 教授; 主要研究方向:信息系统

宋世民   男, 高级工程师; 主要研究方向:卫星综合测试

李可   男, 高级实验师; 主要研究方向:环境控制与计算机测试

通讯作者:
李可, E-mail:like@buaa.edu.cn

中图分类号: V221⁺.3;TB553
计量
- 文章访问数: 849
- HTML全文浏览量: 147
- PDF下载量: 376
- 被引次数: 0
出版历程
- 收稿日期: 2016-08-15
- 录用日期: 2016-12-09
- 网络出版日期: 2017-09-20

Multi-classification spacecraft electrical signal identification method based on random forest

LAN Wei¹,
JIA Suling¹,
SONG Shimin²,
LI Ke^{3
, ,}

1.
School of Economics and Management, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
2.
China Academy of Space Technology, Beijing 100194, China
3.
School of Aeronautic Science and Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China

Funds:

Aeronautical Science Foundation of China 2012XX1043

the Fundamental Research Funds for the Central Universities YWF-16-HKXY-017

National Natural Science Foundation of China 61773039

More Information

Corresponding author: LI Ke, E-mail:like@buaa.edu.cn

摘要

摘要:
针对航天器电特性信号数据存在数据量大、特征维数高、计算复杂度大和识别率低等问题，提出基于主成分分析（PCA）的特征提取方法和随机森林（RF）算法，对原始数据进行降维，提高计算效率和识别率，实现对航天器电信号数据的快速、准确识别分类。随机森林算法在处理高维数据上具有优越的性能，但是考虑到时间复杂度问题，利用主成分分析方法对数据进行压缩和降维，在保证准确率的同时提高了计算效率。实验结果表明：与其他算法相比，针对航天器电特性信号数据，本文方法在准确率、计算效率和稳定性等方面均显示出优异的性能。
- 航天器 /
- 电信号识别 /
- 主成分分析(PCA) /
- 多分类 /
- 随机森林(RF)
Abstract:
The spacecraft electrical signal characteristic data have problems such as large amount, high-dimensional features, high computational complexity and low identification rate. The feature extraction method of principal component analysis (PCA) and random forest (RF) algorithm was proposed to reduce the dimensionality of the original data, improve the computational efficiency and identification rate, and achieve rapid and accurate identification of spacecraft electrical signal data. The random forest algorithm has superior performance in dealing with high-dimensional data. However, considering the time complexity, the method of PCA was used to compress the data and reduce the dimension in order to ensure the accuracy of the classification and improve the computational efficiency. The experimental results show that compared with other algorithms, the proposed method shows excellent performance in accuracy, computational efficiency, and stability when dealing with spacecraft electrical signal data.
- spacecraft /
- electrical signal identification /
- principal component analysis (PCA) /
- multi-classification /
- random forest (RF)

HTML全文

图 1 本文算法设计流程

Figure 1. Design flowchart of proposed algorithm

下载: 全尺寸图片幻灯片

图 2 PCA方法流程

Figure 2. Flowchart of PCA method

下载: 全尺寸图片幻灯片

图 3 RF算法流程

Figure 3. Flowchart of RF algorithm

下载: 全尺寸图片幻灯片

图 4 决策树结构

Figure 4. Structure of decision tree

下载: 全尺寸图片幻灯片

图 5 部分电特性数据的物理意义

Figure 5. Physical meaning of some electrical characteristic data

下载: 全尺寸图片幻灯片

图 6 决策树数目与分类误差率曲线

Figure 6. Curves for number of decision trees and classification error rate

下载: 全尺寸图片幻灯片

表 1 训练时间和预测准确率对比

Table 1. Comparison of training time and prediction accuracy

算法	准确率/%	训练时间/s
NBM	79.02
KNN	85.43	127.36
SVM	88.23	1 873.80
RF	98.90	189.93
PCA-NBM	81.41
PCA-KNN	94.34	11.33
PCA-SVM	91.59	29.32
PCA-RF	98.33	36.40

下载: 导出CSV

参考文献(17)

[1]	魏传锋, 贾阳, 王浚.航天器在轨自主热故障诊断专家系统研究[J].装备环境工程, 2006, 3(3):54-57. http://www.cnki.com.cn/Article/CJFDTOTAL-JSCX200603012.htm WEI C F, JIA Y, WANG J.Research on in-orbit spacecraft thermal fault diagnosis expert system[J].Equipment Environmental Engineering, 2006, 3(3):54-57(in Chinese). http://www.cnki.com.cn/Article/CJFDTOTAL-JSCX200603012.htm
[2]	SHAW S R.System identification techniques and modeling for nonintrusive load diagnostics[D].Cambrige:Massachusetts Institute of Technology, 2000.
[3]	李可.多参数环境模拟系统的智能控制方法与仿真研究[J].北京航空航天大学学报, 2007, 33(5):535-538. http://bhxb.buaa.edu.cn/CN/abstract/abstract9546.shtml LI K.System model simulation and control method used in environmental simulation chambers[J].Journal of Beijing University of Aeronautics and Astronautics, 2007, 33(5):535-538(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract9546.shtml
[4]	LUO R.Analysis of PHM technology for spacecraft[J].Spacecraft Engineering, 2013, 22(4):95-102.
[5]	LIU Y, LI K, HUANG Y.Spacecraft electrical characteristics identification study based on offline FCM clustering and online SVM classifier[C]//International Conference on Multisensor Fusion and Information Integration for Intelligent Systems (MFI).Piscataway, NJ:IEEE Press, 2014:1-4.
[6]	LI K, LIU Y, WANG Q.A spacecraft electrical characteristics multi-label classification method based on off-line FCM clustering and on-line WPSVM[J].Plos One, 2015, 10(11):1413-1423.
[7]	李可, 刘祎, 杜少毅.基于PCA和WPSVM的航天器电特性识别方法[J].北京航空航天大学学报, 2015, 41(7):1177-1182. http://bhxb.buaa.edu.cn/CN/abstract/abstract13308.shtml LI K, LIU Y, DU S Y.Spacecraft electrical characteristics identification method based on PCA feature extraction and WPSVM[J].Journal of Beijing University of Aeronautics and Astronautics, 2015, 41(7):1177-1182(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract13308.shtml
[8]	鄢仁武, 叶轻舟, 周理.基于随机森林的电力电子电路故障诊断技术[J].武汉大学学报(工学版), 2013, 46(6):742-746. http://www.cnki.com.cn/Article/CJFDTOTAL-WSDD201306012.htm YAN R W, YE Q Z, ZHOU L.Application of random forests algorithm to fault diagnosis of power electronic circuit[J].Engineering Journal of Wuhan University, 2013, 46(6):742-746(in Chinese). http://www.cnki.com.cn/Article/CJFDTOTAL-WSDD201306012.htm
[9]	庄进发, 罗键, 彭彦卿, 等.基于改进随机森林的故障诊断方法研究[J].计算机集成制造系统, 2009, 15(4):777-785. http://www.cnki.com.cn/Article/CJFDTOTAL-JSJJ200904026.htm ZHUANG J F, LUO J, PENG Y Q, et al.Fault diagnosis method based on modified random forests[J].Computer Integrated Manufacturing Systems, 2009, 15(4):777-785(in Chinese). http://www.cnki.com.cn/Article/CJFDTOTAL-JSJJ200904026.htm
[10]	LI K, LIU W K, WANG J, et al.An intelligent control method for a large multi-parameter environmental simulation cabin[J].Chinese Journal of Aeronautics, 2013, 26(6):1360-1369. doi: 10.1016/j.cja.2013.07.006
[11]	LI K, LIU W K, WANG J, et al.Multi-parameter decoupling and slope tracking control strategy of a large-scale high altitude environment simulation test cabin[J].Chinese Journal of Aeronautics, 2014, 27(6):1390-1400. doi: 10.1016/j.cja.2014.10.005
[12]	LIU Y, LI K, SONG S M, et al.The research of spacecraft electrical characteristics identification and diagnosis using PCA feature extraction[C]//IEEE International Conference on Signal Processing.Piscataway, NJ:IEEE Press, 2014:1413-1417.
[13]	刘小虎, 李生.决策树的优化算法[J].软件学报, 1998, 9(10):797-800. http://cdmd.cnki.com.cn/Article/CDMD-10732-1014421484.htm LIU X H, LI S.Optimization algorithm of decision tree[J].Journal of Software, 1998, 9(10):797-800(in Chinese). http://cdmd.cnki.com.cn/Article/CDMD-10732-1014421484.htm
[14]	PATAKI B, TOTH N.Classification confidence weighted majority voting using decision tree classifiers[J].International Journal of Intelligent Computing & Cybernetics, 2008, 1(2):169-192. https://www.deepdyve.com/lp/emerald-publishing/classification-confidence-weighted-majority-voting-using-decision-tree-Q2YteXnQ3F
[15]	PAL M.Random forest classifier for remote sensing classification[J].International Journal of Remote Sensing, 2005, 26(1):217-222. doi: 10.1080/01431160412331269698
[16]	DENG H, RUNGER G.Gene selection with guided regularized random forest[J].Pattern Recognition, 2013, 46(12):3483-3489. doi: 10.1016/j.patcog.2013.05.018
[17]	KHAING H K T.Detection model for daniel-of-service attacks using random forest and k-nearest neighbors[J].International Journal of Advanced Research in Computer Engineering & Technology, 2013, 2(5):1855-1860. http://www.ijpttjournal.org/volume-3/issue-1/IJPTT-V3I1P413.pdf