吉林大学学报(地球科学版) ›› 2016, Vol. 46 ›› Issue (2): 563-568.doi: 10.13278/j.cnki.jjuese.201602206

• 地质工程与环境工程 • 上一篇    下一篇

基于整体经验模态分解和支持向量回归的北京市PM2.5预测

秦喜文1,2,3, 刘媛媛2, 王新民2, 董小刚2, 张瑜2, 周红梅2   

  1. 1. 长春工业大学研究生院, 长春 130012;
    2. 长春工业大学基础科学学院, 长春 130012;
    3. 长春工业大学汽车工程研究院, 长春 130012
  • 收稿日期:2015-07-01 发布日期:2016-03-26
  • 通讯作者: 王新民(1957-),男,教授,主要从事数值分析方面的研究,E-mail:wxm@jlu.edu.cn E-mail:wxm@jlu.edu.cn
  • 作者简介:秦喜文(1979-),男,副教授,主要从事HHT理论与应用方面的研究,E-mail:qinxiwen@ccut.edu.cn
  • 基金资助:

    国家自然科学基金项目(11301036,11226335,51278065);吉林省教育厅科研项目(2014第127号,2013第142号)

PM2.5 Prediction of Beijing City Based on Ensemble Empirical Mode Decomposition and Support Vector Regression

Qin Xiwen1,2,3, Liu Yuanyuan2, Wang Xinmin2, Dong Xiaogang2, Zhang Yu2, Zhou Hongmei2   

  1. 1. Graduate School, Changchun University of Technology, Changchun 130012, China;
    2. School of Basic Sciences, Changchun University of Technology, Changchun 130012, China;
    3. Automotive Engineering Research Institute, Changchun University of Technology, Changchun 130012, China
  • Received:2015-07-01 Published:2016-03-26
  • Supported by:

    Supported by National Natural Science Foundation of China(11301036,11226335,51278065)and Scientific Research Project of Jilin Province Department of Education(No.127 in 2014,No.142 in 2013)

摘要:

为了更好地掌握大气中PM2.5浓度的变化规律,利用EEMD-SVR混合模型对该地区的PM2.5浓度值进行了短期预测。首先,通过采用整体经验模态分解(EEMD)方法分析北京市PM2.5,把原始时间序列分解成多个固有模态函数和趋势项;然后,对各阶固有模态函数进行周期性分析,揭示了北京市PM2.5的周期性变化特点;最后,对经过EEMD分解后的各阶固有模态函数和趋势项用支持向量机回归(SVR)方法进行预测。结果表明, EEMD-SVR混合模型比单一的SVR模型预测精度更高。

关键词: 整体经验模态分解, 固有模态函数, 周期性, 支持向量机回归

Abstract:

In order to obtain the pattern of variation of PM2.5 concentrations in the atmosphere in Beijing City, we build a EEMD-SVR hybrid model that can predict the PM2.5 level in a short term. Firstly, according to the ensemble empirical mode decomposition (EEMD) method to analyse the PM2.5 of Beijing City, the original time series is decomposed into the series of intrinsic mode functions (IMFs) and trend items; then, the periodic variation characteristics of PM2.5 is revealed through the periodic analysis of each intrinsic mode function;finally, we use support vector regression (SVR) to forecast all IMFs and trend items, which reflect the rationality of using SVR model. The results show that the prediction accuracy of mixed EEMD-SVR model is higher than single SVR model.

Key words: ensemble empirical mode decomposition, intrinsic mode functions (IMF), periodicity, support vector regression

中图分类号: 

  • C81

[1] 刘贺,张弘强.基于粒子群优化神经网络算法的深基坑变形预测方法[J].吉林大学学报(地球科学版),2014,44(5):1609-1614. Liu He, Zhang Hongqiang. A Prediction Method for the Deformation of Deep Foundation Pit Based on the Particle Swarm Optimization Neural Network[J]. Journal of Jilin University(Earth Science Edition), 2014,44(5):1609-1614.

[2] 蒋玲玲,熊德琪,张新宇.大连滨海湿地景观格局变化及其驱动机制[J].吉林大学学报(地球科学版),2008,38(4):673-674. Jiang Lingling, Xiong Deqi, Zhang Xinyu. Change of Landscape Pattern and Its Driving Mechanism of the Coastal Wetland in Dalian City[J].Journal of Jilin University(Earth Science Edition),2008,38(4):673-674.

[3] 董志颖,李兵,孙晶.GIS支持下的吉林西部水质预警系统[J].吉林大学学报(地球科学版),2003,33(1):56-58. Dong Zhiying, Li Bing, Sun Jing. The Research of Forecast of Water Quality in the Western Part of Jilin Province by Means of GIS[J].Journal of Jilin University(Earth Science Edition),2003,33(1):56-58.

[4] 潘保芝, 石玉江, 蒋必辞.致密砂岩气层压裂产能及等级预测方法[J]. 吉林大学学报(地球科学版), 2015, 45(2):649-654. Pan Baozhi, Shi Yujiang, Jiang Bici.Research on Gas Yield and Level Predition for Post-Frac Tight Sandstone Reservoirs[J]. Journal of Jilin University(Earth Science Edition), 2015, 45(2):649-654.

[5] 张艺耀,苗冠鸿.影响PM2.5因素的多元统计分析与预测[J].资源节约与环保,2013(11):13-16. Zhang Yiyao, Miao Guanhong. The Factors Affecting PM2.5 and PM2.5 Forecasting Based on Multivariate Statistical Analysis[J].Resource Economization & Environment Protection, 2013(11):13-16.

[6] 张怡文,胡静宜,王冉.基于神经网络的PM2.5预测模型研究[J].江苏师范大学学报(自然科学版),2015, 33(1):63-65. Zhang Yiwen, Hu Jingyi, Wang Ran. PM2.5 Prediction Model Based on Neural Network[J].Journal of Jiangsu Normal University (Natural Science Edition), 2015, 33(1):63-65.

[7] 王敏, 邹滨, 郭宇. 基于BP人工神经网络的城市PM2.5浓度空间预测[J].环境污染与防治,2013,35(9):63-70. Wang Min, Zou Bin, Guo Yu. BP Artificial Neural Network-Based Analysis of Spatial Variability of Urban PM2.5 Concentration[J].Environmental Pollution & Control,2013,35(9):63-70.

[8] Zhou Qingping, Jiang Haiyan. A Hybrid Model for PM2.5 Forecasting Based on Ensemble Empirical Mode Decomposition and a General Gegression Neural Network[J]. Science of the Total Environment,2014, 496:264-274.

[9] Huang N E,Shen Z. The Empirical Mode Decomposition and Hillbert Spectrum for Nonlinear and Non-stationary Time Series Analysis[J]. Proceedings of the Royal Society London, 1998,454:903-995.

[10] Wu Zhaohua,Huang Norden E.A Study of the Ch-aracteristics of White Noise Using the Empirical Mode Decomposition Method[J].Proceedings of the Royal Society,2004, 460:1597-1611.

[11] Vapnik V. The Nature of Statistical Learning Theory[M]. New York:Springer-Verlag, 1995.

[12] 刘子阳,郭崇慧.应用支持向量回归方法预测胎儿体重[D].大连:大连理工大学,2005. Liu Ziyang, Guo Chonghui. Fetal Weight Prediction by Using Support Vector Regression[D].Dalian:Dalian University of Technology,2005.

[13] 范瑜,邹塞.徐州市春季PM10及PM2.5污染来源分析[J].环境科技,2014,27(2):49-52. Fan Yu, Zou Sai.Analysis of the PM10& PM2.5 Pollution Sources of Xuzhou in Spring[J].Environmental Science and Technology, 2014,27(2):49-52.

[14] 蔡赟姝,卢志明.基于经验模态分解的上证综合指数时间序列分析[J].上海大学学报(自然科学版),2012,18(4):384-389. Cai Yunshu, Lu Zhiming.The Shanghai Composite Index Time Series Analysis Based on Empirical Mode Decomposition[J].Journal of Shanghai University(Natural Science Edition),2012,18(4):384-389.

[1] 王洁, 宫辉力, 陈蓓蓓, 高明亮, 周超凡, 梁悦, 陈文锋. 基于Morlet小波技术的北京平原地面沉降周期性分析[J]. 吉林大学学报(地球科学版), 2018, 48(3): 836-845.
[2] 董烈乾, 李振春, 刘磊, 李志娜, 桑运云. 基于经验模态分解的曲波阈值去噪方法[J]. J4, 2012, 42(3): 838-844.
[3] 葸晓宇,刘 洪. HHT方法在研究地震旋回体中的应用[J]. J4, 2007, 37(3): 624-0628.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!