基于ECOC平衡随机森林的雷达降水粒子分类

doi:10.12305/j.issn.1001-506X.2024.05.14

系统工程与电子技术 ›› 2024, Vol. 46 ›› Issue (5): 1599-1606.doi: 10.12305/j.issn.1001-506X.2024.05.14

• 传感器与信号处理 • 上一篇

基于ECOC平衡随机森林的雷达降水粒子分类

李海¹^,*, 田众¹, 钱君²

1. 中国民航大学天津市智能信号与图像处理重点实验室, 天津 300300
2. 中国航空工业集团公司雷华电子技术研究所, 江苏无锡 214063

收稿日期:2022-12-31 出版日期:2024-04-30 发布日期:2024-04-30
通讯作者: 李海
作者简介:李海(1976—), 男, 教授, 博士, 博士研究生导师, 主要研究方向为机载气象雷达信号处理及机器学习在气象雷达中的应用、分布式目标检测与参数估计、自适应信号处理
田众(1997—), 男, 硕士研究生, 主要研究方向为机载气象雷达信号处理、机器学习在气象雷达中的应用
钱君(1985—), 男, 高级工程师, 学士, 主要研究方向为机载气象雷达系统
基金资助:
民机项目(MJ-2018-S-28);天津市自然基金重点项目(20JCZDJC00490);中央高校基本科研业务经费中国民航大学专项(3122019041)

Hydrometeor classification for radar based on ECOC-balanced random forest

Hai LI¹^,*, Zhong TIAN¹, Jun QIAN²

1. Tianjin Key Lab for Advanced Signal Processing, Civil Aviation University of China, Tianjin 300300, China
2. Leihua Electronic Technology Research Institute, Aviation Industry Corporation of China, Wuxi 214063, China

Received:2022-12-31 Online:2024-04-30 Published:2024-04-30
Contact: Hai LI

摘要/Abstract

摘要：

针对数据不平衡情况下的降水粒子分类问题, 提出了一种基于纠错输出码(error correcting output code, ECOC)平衡随机森林的双偏振气象雷达降水粒子分类方法。首先, 将多类别降水粒子数据集编码为多个二分类数据集; 然后, 对二分类数据集进行有放回的平衡重采样, 构建多棵分类回归树; 最后, 利用所有的分类回归树联合进行降水粒子分类。对实测数据的处理结果表明, 所提方法能够在保证总体准确率较高的情况下, 大幅提高少数类的分类效果。

关键词: 双偏振气象雷达, 降水粒子分类, 数据不平衡, 纠错输出码, 平衡随机森林

Abstract:

To address the problem of hydrometeor classification with data imbalance condition, this paper proposes a hydrometeor classification method based on error correcting output code (ECOC) balanced random forest for dual-polarization weather radar. Firstly, the multiclass hydrometeor dataset is coded into multiple binary datasets, and then the binary datasets are balanced resampling with replacement to construct multiple classification and regression trees. Finally, all the classification and regression trees are used to jointly classify hydrometeors. The processing results of the measured data indicate that the proposed method can significantly improve the classification effect of minority classes while ensuring a high overall accuracy.

Key words: dual-polarization weather radar, hydrometeor classification, data imbalance, error correcting output code (ECOC), balanced random forest

中图分类号:

959.4

李海, 田众, 钱君. 基于ECOC平衡随机森林的雷达降水粒子分类[J]. 系统工程与电子技术, 2024, 46(5): 1599-1606.

Hai LI, Zhong TIAN, Jun QIAN. Hydrometeor classification for radar based on ECOC-balanced random forest[J]. Systems Engineering and Electronics, 2024, 46(5): 1599-1606.

图/表 11

图1

图2

表1

图3

表2

表3

图4

表4

图5

表5

表6

参考文献 31

1	LISCHI S, LUPIDI A, BERIZZI F, et al. X-band full polarized Doppler weather radar return simulation by using propagation-modified ensemble-averaged covariance matrix[C]//Proc. of the International Radar Symposium. Dresden IEEE Press, 2013.
2	LISCHI S, LUPIDI A, MARTORELLA M, et al. Advanced polarimetric Doppler weather radar simulator[C]//Proc. of the 15th International IEEE Radar Symposium, 2014.
3	ZHANG G . Weather radar polarimetry[M]. Oxford: CRC Press, Inc., 2016.
4	LI Z , ZHANG Y , ZHANG G F , et al. A microphysics-based si-mulator for advanced airborne weather radar development[J]. IEEE Trans.on Geoscience and Remote Sensing, 2011, 49 (4): 1356- 1373. doi: 10.1109/TGRS.2010.2076354
5	LI Z , PERERA S , ZHANG Y , et al. Phased-array radar system simulator (PASIM): development and simulation result assessment[J]. Remote Sensing, 2019, 11 (4): 422. doi: 10.3390/rs11040422
6	BRINGI V N , CHANDRASEKAR V . Polarimetric Doppler weather radar: principles and applications[M]. Carnbridge: Cambridge University Press, 2001.
7	李海, 白锦, 孙研, 等. 基于修正小波变换插值-TAN的雷达降水粒子分类[J]. 系统工程与电子技术, 2022, 44 (5): 1527- 1535. doi: 10.12305/j.issn.1001-506X.2022.05.13
	LI H , BAI J , SUN Y , et al. Radar hydrometeor classification based on modified wavelet transform interpolation-TAN[J]. Systems Engineering and Electronics, 2022, 44 (5): 1527- 1535. doi: 10.12305/j.issn.1001-506X.2022.05.13
8	覃文娜. 江苏省冰雹发生规律及数值模拟技术研究[D]. 南京: 南京信息工程大学, 2012.
	TAN W N. Hail occurrence regularity and numerical simulation technology in Jiangsu province[D]. Nanjing: Nanjing University of Information Science and Technology, 2012.
9	LI H , FENG K H , GUO M H . Echo simulation of non-spherical ice crystal particles for airborne weather radar[J]. IET Radar, Sonar Navigation, 2022, 17 (2): 277- 294.
10	STRAKA J M. Hydrometeor fields in a supercell storm as deduced from dual-polarization radar[C]//Proc. of the 18th Conference on Severe Local Storms, 1996: 551-554.
11	HYANGSUK P , RYZHKOV A V , ZRNI D S , et al. The hydrometeor classification algorithm for the polarimetric WSR-88D: description and application to an MCS[J]. Weather and Forecasting, 2009, 24 (3): 730- 748. doi: 10.1175/2008WAF2222205.1
12	HASSAN A S , BOUMAHMOUD A A , BEATRICE F , et al. A new fuzzy logic hydrometeor classification scheme applied to the french X-band, C-band, and S-band polarimetric radars[J]. Journal of Applied Meteorology and Climatology, 2013, 52 (10): 2328- 2344. doi: 10.1175/JAMC-D-12-0236.1
13	MARZANO F S , SCARANARI D , CELANO M , et al. Hydrometeor classification from dual-polarized weather radar: extending fuzzy logic from S-band to C-band data[J]. Advances in Geosciences, 2006, 7 (6): 109- 114.
14	冯亮, 肖辉, 孙跃. X波段双偏振雷达水凝物粒子相态识别应用研究[J]. 气候与环境研究, 2018, 23 (3): 366- 386.
	FENG L , XIAO H , SUN Y . A study on hydrometeor classification and application based on X-band dual-polarization radar measurements[J]. Climatic and Environmental Research, 2018, 23 (3): 366- 386.
15	RAUT B A , KONWAR M , MURUGAVEL P , et al. Microphysical origin of raindrop size distributions during the Indian monsoon[J]. Geophysical Research Letters, 2021, 48 (16) doi: 10.1029/2021GL093581
16	CHEN Y C , LIU X E , BI K , et al. Hydrometeor classification of winter precipitation in northern china based on multiplatform radar observation system[J]. Remote Sensing, 2021, 13 (24): 5070- 5078. doi: 10.3390/rs13245070
17	GRAZIOLI J , TUIA D , BERNE A . Hydrometeor classification from polarimetric radar measurements: a clustering approach[J]. Atmospheric Measurement Techniques, 2015, 8 (1): 149- 170. doi: 10.5194/amt-8-149-2015
18	杨淑群, 芮景析, 冯汉中. 支持向量机(SVM)方法在降水分类预测中的应用[J]. 西南农业大学学报(自然科学版), 2006, 28 (2): 252- 257. doi: 10.3969/j.issn.1673-9868.2006.02.020
	YANG S Q , RUI J X , FENG H Z . Application of support vector machine (SVM) in rainfall categorical forecast[J]. Journal of Southwest Agricultural University(Natural Science), 2006, 28 (2): 252- 257. doi: 10.3969/j.issn.1673-9868.2006.02.020
19	李海, 尚金雷, 孙婷逸, 等. 一种基于离散属性BNT的双偏振气象雷达降水粒子分类方法[J]. 电子学报, 2021, 49 (3): 619- 624.
	LI H , SHANG J L , SUN T Y , et al. A BNT hydrometeor classification algorithm for dual-polarization radar[J]. Acta Electronica Sinica, 2021, 49 (3): 619- 624.
20	MARZANO F S, SCARANARI D, MONTOPOLI M, et al. Model-oriented hydrometeor classification and water content estimate using dual-polarized weather radars[C]//Proc. of the Radar Conference, 2008.
21	YANG J , KUN Z , ZHANG G , et al. A Bayesian hydrometeor classification algorithm for C-band polarimetric radar[J]. Remote Sensing, 2019, 11 (16): 1884- 1894. doi: 10.3390/rs11161884
22	BECHINI R , CHANDRASEKAR V . A semisupervised robust hydrometeor classification method for dual-polarization radar applications[J]. Journal of Atmospheric and Oceanic Technology, 2015, 32 (1): 22- 47. doi: 10.1175/JTECH-D-14-00097.1
23	HAIBO H, YANG B, EDWARDO A G, et al. Adasyn: adaptive synthetic sampling approach for imbalanced learning[C]//Proc. of the IEEE International Joint Conference on Neural Networks, 2008: 1322-1328.
24	LIU X Y , WU J X , ZHOU Z H . Exploratory undersampling for class-imbalance learning[J]. IEEE Trans.on Systems, Man, and Cybernetics. Part B, Cybernetics: a publication of the IEEE Systems, Man, and Cybernetics Society, 2009, 39 (2): 539- 550. doi: 10.1109/TSMCB.2008.2007853
25	ELREEDY D , ATIYA A F . A comprehensive analysis of synthetic minority oversampling technique (SMOTE) for handling class imbalance[J]. Information Sciences, 2019, 505, 32- 64. doi: 10.1016/j.ins.2019.07.070
26	HIEN M N , ERIC W C , KATSUARI K . Borderline over-sampling for imbalanced data classification[J]. International Journal of Knowledge Engineering and Soft Data Paradigms, 2011, 3 (1): 4- 21. doi: 10.1504/IJKESDP.2011.039875
27	PRINCE M , PRATHAP J M P . An imbalanced dataset and class overlapping classification model for big data[J]. Computer Systems Science and Engineering, 2023, 44 (2): 1009- 1024. doi: 10.32604/csse.2023.024277
28	GARCIA V , SANCHEZ J S , MOLLINEDA R A . On the effectiveness of preprocessing methods when dealing with differ-ent levels of class imbalance[J]. Knowledge-Based Systems, 2012, 25 (1): 13- 21. doi: 10.1016/j.knosys.2011.06.013
29	PRASETIYOWATI M I , MAULIDEVI N U , SURENDRO K . The accuracy of random forest performance can be improved by conducting a feature selection with a balancing strategy[J]. PeerJ. Computer science, 2022, 8, e1041. doi: 10.7717/peerj-cs.1041
30	ASIM Y , MALIK A K , RAZA B , et al. Predicting influential blogger's by a novel, hybrid and optimized case based reasoning approach with balanced random forest using imbalanced data[J]. IEEE Access, 2021, 9, 6836- 6854. doi: 10.1109/ACCESS.2020.3048610
31	GUPTA S C , GOEL N . Predictive modeling and analytics for diabetes using hyperparameter tuned machine learning techniques[J]. Procedia Computer Science, 2023, 218, 1257- 1269. doi: 10.1016/j.procs.2023.01.104

超参数类别	超参数初值
平衡随机森林数量M	(30, 35, 40, 45, 50)
每个平衡随机森林包含的CART树数量L	(10, 50, 100, 200, 300)
CART树每次分裂抽取的特征数量P	(1, 2, 3)

类标签	类别
C=0	冰晶
C=1	雪
C=2	雨
C=3	霰
C=4	冰雹

类别	数据集
类别	1	2	3	4	…	28	29	30
冰晶	0	1	1	1	…	1	0	1
雪	0	0	1	0	…	0	1	0
雨	1	1	0	0	…	1	0	1
霰	1	1	1	1	…	0	1	0
冰雹	0	0	1	1	…	1	0	1

	粒子个数	粒子占比/%
冰晶	3 080	2.51
雪	27 754	22.61
雨	86 613	70.56
霰	2 458	2.00
冰雹	2 847	2.32

真实情况(真实值)	分类结果(预测值)
真实情况(真实值)	正例	反例
正例	真正例(true positive, TP)	假反例(false negative, FN)
反例	假正例(false positive, FP)	真反例(true negative, TN)

基于ECOC平衡随机森林的雷达降水粒子分类

Hydrometeor classification for radar based on ECOC-balanced random forest

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 31

相关文章 1

编辑推荐

Metrics

本文评价

算法	准确率	F1度量	平均召回率
SVM	79.29	52.12	46.56
欠采样SVM	53.75	41.81	73.95
ECOC平衡随机森林	74.00	59.50	78.78