深圳市区空气污染的人工神经网络预测
|
摘要
利用深圳市2006至2013年的大气污染物监测浓度数据和气象资料,分析深圳市空气质量的逐月分布变化特征。采用Pearson相关分析,选择显著相关因子,分别以BP神经网络和RBF神经网络构建空气质量预测模型,对该市2013年SO2、NO2、PM103种空气污染物的月均值进行预测。实验结果表明,通过Pearson相关分析建立的预测模型有更高的预报精度。BP和RBF 2种网络预测效果都比较理想,对不同污染物的预测精度各有高低。但BP网络的构建和参数优化过程较为复杂且网络训练结果不稳定,而RBF网络构建和训练简单,时间短而结果稳定。在综合性能上,RBF网络用于环境空气污染物浓度的预测具有更强的适用性。
The air pollutants concentrations and meteorological data monitored from 2006 to 2013 were used to analyze the monthly distribution characteristics of air quality in Shenzhen.On the basis of applying the Pearson correlation analysis to select significant correlation factors,air quality prediction models were built by the BP and RBF neural network,respectively,and the monthly average concentrations of SO2,NO2 and PM10were forecasted in 2013.The results are demonstrated that the prediction model established by the Pearson correlation analysis has higher prediction accuracy.Both BP and RBF network are ideal in predict effect,and each network has different prediction accuracy for different pollutants.The processes of establishment and parameter optimization using the BP network are complicated,and the results of network training are not stable.Compared to the establishment and training of the BP network,the RBF network is more simple,and the results are more stable.On comprehensive performance,the RBF network has better applicability in prediction of air pollutants concentrations.
The air pollutants concentrations and meteorological data monitored from 2006 to 2013 were used to analyze the monthly distribution characteristics of air quality in Shenzhen.On the basis of applying the Pearson correlation analysis to select significant correlation factors,air quality prediction models were built by the BP and RBF neural network,respectively,and the monthly average concentrations of SO2,NO2 and PM10were forecasted in 2013.The results are demonstrated that the prediction model established by the Pearson correlation analysis has higher prediction accuracy.Both BP and RBF network are ideal in predict effect,and each network has different prediction accuracy for different pollutants.The processes of establishment and parameter optimization using the BP network are complicated,and the results of network training are not stable.Compared to the establishment and training of the BP network,the RBF network is more simple,and the results are more stable.On comprehensive performance,the RBF network has better applicability in prediction of air pollutants concentrations.
引文
[1]Jacob D.J.,Winner D.A.Effect of climate change on air quality.Atmospheric Environment,2009,43(1):51-56
[2]Kumar A.,Goyal P.Forecasting of air quality index in delhi using neural network based on principal component analysis.Pure and Applied Geophysics,2013,170(4):711-722
[3]赵宏,刘爱霞,王恺,等.环境空气SO2和NO2浓度的GA_ANN预测模型研究.计算机工程与应用,2010,46(8):199-201,213Zhao Hong,Liu Aixia,Wang Kai,et al.Study on GA_ANN based ambient air SO2and NO2concentration forecasting model.Computer Engineering and Applications,2010,46(8):199-201,213(in Chinese)
[4]Cobaner M.,Citakoglu H.,KisiO.,et al.Estimation of mean monthly air temperatures in Turkey.Computers and Electronics in Agriculture,2014,109:71-79
[5]Westerlund J.,Urbain J.P.,Bonilla J.Application of air quality combination forecasting to Bogota.Atmospheric Environment,2014,89:22-28
[6]'Ciric'I.T.,C'ojbaic'.M.,Nikolic'V.D.,et al,Air quality estimation by computational intelligence methodologies.Thermal Science,2012,16(Suppl 2):S493-S504
[7]Muoz E.,Martín M.L.,Turias I.J.,et al.Prediction of PM10and SO2exceedances to control air pollution in the Bay of Algeciras,Spain.Stochastic Environmental Research and Risk Assessment,2014,28(6):1409-1420
[8]孙大洪,王发展,刘强,等.基于BP和RBF神经网络的滚动轴承故障诊断比较.轴承,2010,(2):53-56Sun Dahong,Wang Fazhan,Liu Qiang,et al.Comparative study on fault diagnosis of rolling bearings based on BP and RBF neural network.Bearing,2010,(2):53-56(in Chinese)
[9]白鹤鸣,沈润平,师华定,等.基于BP神经网络的空气污染指数预测模型研究.环境科学与技术,2013,36(3):186-189Bai Heming,Shen Runping,Shi Huading,et al.Forecasting model of air pollution index based on BP neural network.Environmental Science&Technology,2013,36(3):186-189(in Chinese)
[10]刘杰,杨鹏,吕文生,等.城市空气质量的BP和RBF人工神经网络建模及分类评价.安全与环境工程,2014,21(6):129-134,139Liu Jie,Yang Peng,Lu Wensheng,et al.BP and RBF artificial neural network modeling and classified evaluation for urban air quality.Safety and Environmental Engineering,2014,21(6):129-134,139(in Chinese)
[11]朱树先,张仁杰.BP和RBF神经网络在人脸识别中的比较.仪器仪表学报,2007,28(2):375-379Zhu Shuxian,Zhang Renjie.Comparison with BP and RBF neural network used in face recognition.Chinese Journal of Scientific Instrument,2007,28(2):375-379(in Chinese)
[12]高彩云,高宁.变形预报中RBF和BP神经网络有效性比较.人民黄河,2012,34(7):132-134Gao Caiyun,Gao Ning.Comparison with RBF and BP neural network used in deformation prediction.Yellow River,2012,34(7):132-134(in Chinese)
[13]深圳市气象局网站.http://www.szmb.gov.cn/
[14]谭德生,陈军苗.2004~2009年深圳空气质量及其污染物扩散条件分析.南昌大学学报(理科版),2012,36(3):301-306Tan Desheng,Chen Junmiao.Analyse of air quality and diffusive conditions of contaminant in Shenzhen from 2004to 2009.Journal of Nanchang University(Natural Science),2012,36(3):301-306(in Chinese)
[15]朱星宇,陈勇强.SPSS多元统计分析方法及应用.北京:清华大学出版社,2011
[16]刘国红,彭朝琼,黄广文,等.深圳市2002~2005年大气污染物浓度与气象因素的关系研究.现代预防医学,2008,35(15):2855-2857Liu Guohong,Peng Zhaoqiong,Huang Guangwen,et al.Study on relationship between concentrations of air pollutants and meteorological factor in Shenzhen from 2002 to2005.Modern Preventive Medicine,2008,35(15):2855-2857(in Chinese)
[17]陈明.MATLAB神经网络原理与实例精解.北京:清华大学出版社,2013
[18]张德丰.MATLAB神经网络应用设计.北京:机械工业出版社,2012
[2]Kumar A.,Goyal P.Forecasting of air quality index in delhi using neural network based on principal component analysis.Pure and Applied Geophysics,2013,170(4):711-722
[3]赵宏,刘爱霞,王恺,等.环境空气SO2和NO2浓度的GA_ANN预测模型研究.计算机工程与应用,2010,46(8):199-201,213Zhao Hong,Liu Aixia,Wang Kai,et al.Study on GA_ANN based ambient air SO2and NO2concentration forecasting model.Computer Engineering and Applications,2010,46(8):199-201,213(in Chinese)
[4]Cobaner M.,Citakoglu H.,KisiO.,et al.Estimation of mean monthly air temperatures in Turkey.Computers and Electronics in Agriculture,2014,109:71-79
[5]Westerlund J.,Urbain J.P.,Bonilla J.Application of air quality combination forecasting to Bogota.Atmospheric Environment,2014,89:22-28
[6]'Ciric'I.T.,C'ojbaic'.M.,Nikolic'V.D.,et al,Air quality estimation by computational intelligence methodologies.Thermal Science,2012,16(Suppl 2):S493-S504
[7]Muoz E.,Martín M.L.,Turias I.J.,et al.Prediction of PM10and SO2exceedances to control air pollution in the Bay of Algeciras,Spain.Stochastic Environmental Research and Risk Assessment,2014,28(6):1409-1420
[8]孙大洪,王发展,刘强,等.基于BP和RBF神经网络的滚动轴承故障诊断比较.轴承,2010,(2):53-56Sun Dahong,Wang Fazhan,Liu Qiang,et al.Comparative study on fault diagnosis of rolling bearings based on BP and RBF neural network.Bearing,2010,(2):53-56(in Chinese)
[9]白鹤鸣,沈润平,师华定,等.基于BP神经网络的空气污染指数预测模型研究.环境科学与技术,2013,36(3):186-189Bai Heming,Shen Runping,Shi Huading,et al.Forecasting model of air pollution index based on BP neural network.Environmental Science&Technology,2013,36(3):186-189(in Chinese)
[10]刘杰,杨鹏,吕文生,等.城市空气质量的BP和RBF人工神经网络建模及分类评价.安全与环境工程,2014,21(6):129-134,139Liu Jie,Yang Peng,Lu Wensheng,et al.BP and RBF artificial neural network modeling and classified evaluation for urban air quality.Safety and Environmental Engineering,2014,21(6):129-134,139(in Chinese)
[11]朱树先,张仁杰.BP和RBF神经网络在人脸识别中的比较.仪器仪表学报,2007,28(2):375-379Zhu Shuxian,Zhang Renjie.Comparison with BP and RBF neural network used in face recognition.Chinese Journal of Scientific Instrument,2007,28(2):375-379(in Chinese)
[12]高彩云,高宁.变形预报中RBF和BP神经网络有效性比较.人民黄河,2012,34(7):132-134Gao Caiyun,Gao Ning.Comparison with RBF and BP neural network used in deformation prediction.Yellow River,2012,34(7):132-134(in Chinese)
[13]深圳市气象局网站.http://www.szmb.gov.cn/
[14]谭德生,陈军苗.2004~2009年深圳空气质量及其污染物扩散条件分析.南昌大学学报(理科版),2012,36(3):301-306Tan Desheng,Chen Junmiao.Analyse of air quality and diffusive conditions of contaminant in Shenzhen from 2004to 2009.Journal of Nanchang University(Natural Science),2012,36(3):301-306(in Chinese)
[15]朱星宇,陈勇强.SPSS多元统计分析方法及应用.北京:清华大学出版社,2011
[16]刘国红,彭朝琼,黄广文,等.深圳市2002~2005年大气污染物浓度与气象因素的关系研究.现代预防医学,2008,35(15):2855-2857Liu Guohong,Peng Zhaoqiong,Huang Guangwen,et al.Study on relationship between concentrations of air pollutants and meteorological factor in Shenzhen from 2002 to2005.Modern Preventive Medicine,2008,35(15):2855-2857(in Chinese)
[17]陈明.MATLAB神经网络原理与实例精解.北京:清华大学出版社,2013
[18]张德丰.MATLAB神经网络应用设计.北京:机械工业出版社,2012