基于BP神经网络预测林内PM_(2.5)浓度
|
摘要
[目的]利用BP神经网络预测林内PM_(2.5)浓度。[方法]利用人工神经网络理论,采用2013年7月—2014年5月野外实时监测数据,建立了以气象参数、污染源强变量和林分结构特征为输入因子,林内PM_(2.5)小时平均浓度为输出因子的预测模型,并对其预测精度进行了评价。[结果]BP人工神经网络模型能够很好地捕捉污染物浓度与气象因素和林分结构间的非线性影响规律,预测结果的平均相对误差为1.71×10~(-3),均方根误差为6.77,拟合优度达0.98,模型具有很高的预测精度。而传统的多元线性回归(MLR)模型预测结果的平均相对误差、均方根误差和拟合优度分别为0.27、22.92和0.93。[结论]研究成果印证了应用BP人工神经网络模型预测林内PM_(2.5)浓度的可行性和准确性。
[Objective]BP neural network was used to predict PM_(2.5)concentration in forest.[Method]An artificial neural network model was established with meteorological data,atmospheric PM_(2.5)concentration outside the forest and forest structure as the input factors,and PM_(2.5)hourly average concentration inside the forest as the output factors. Its prediction accuracy was also evaluated in this paper.[Result]BP artificial neural network model can be trained to model the highly non-linear relationships between meteorological parameters,forest structure and PM_(2.5)concentration.Mean relative error( EMR),root mean square error( E_(RMS)) and goodness of fit( R~2) between BP simulated PM_(2.5)concentrations in the forest and observed ones were 1.71×10~(-3),6.77 and 0.98,respectively.The mean relative error,root mean square error and goodness of fit of traditional MLR model were 0.27,22.92 and 0.93,respectively.[Conclusion]It can be concluded that BP artificial neural network model is a promising approach in predicting PM_(2.5)concentration inside a forest.
[Objective]BP neural network was used to predict PM_(2.5)concentration in forest.[Method]An artificial neural network model was established with meteorological data,atmospheric PM_(2.5)concentration outside the forest and forest structure as the input factors,and PM_(2.5)hourly average concentration inside the forest as the output factors. Its prediction accuracy was also evaluated in this paper.[Result]BP artificial neural network model can be trained to model the highly non-linear relationships between meteorological parameters,forest structure and PM_(2.5)concentration.Mean relative error( EMR),root mean square error( E_(RMS)) and goodness of fit( R~2) between BP simulated PM_(2.5)concentrations in the forest and observed ones were 1.71×10~(-3),6.77 and 0.98,respectively.The mean relative error,root mean square error and goodness of fit of traditional MLR model were 0.27,22.92 and 0.93,respectively.[Conclusion]It can be concluded that BP artificial neural network model is a promising approach in predicting PM_(2.5)concentration inside a forest.
引文
[1]邵龙义,时宗波,黄勤.都市大气环境中可吸入颗粒物的研究[J].环境保护,2000(1):24-26,29.
[2]孙淑萍,古润泽,张晶.北京城区不同绿化覆盖率和绿地类型与空气中可吸入颗粒物(PM10)[J].中国园林,2004,20(3):77-79.
[3]YU Y,SCHLEICHER N,NORRA S,et al.Dynamics and origin of PM2.5during a three-year sampling period in Beijing,China[J].Journal of environmental monitoring,2011,13(2):334-346.
[4]吴海龙,余新晓,师忱,等.PM2.5特征及森林植被对其调控研究进展[J].中国水土保持科学,2012,10(6):116-122.
[5] BECKETT K P,FREER-SMITH P H,TAYLOR G.Particulate pollution capture by urban trees:Effect of species and windspeed[J].Global change biology,2000,6(8):995-1003.
[6] LANGNER M. Reduction of airborne particulates by urban green[J].Bundesamt für naturschutz,2008,179:129-137.
[7]CAVANAGH J A E,ZAWAR-REZA P,WILSON J G.Spatial attenuation of ambient particulate matter air pollution within an urbanised native forest patch[J].Urban forestry&urban greening,2009,8(1):21-30.
[8]郭二果,王成,郄光发,等.北京西山典型游憩林空气悬浮颗粒物季节变化规律[J].东北林业大学学报,2010,38(10):55-57.
[9]殷杉,蔡静萍,陈丽萍,等.交通绿化带植物配置对空气颗粒物的净化效益[J].生态学报,2007,27(11):4590-4595.
[10]汪永英,孙琪,李昭,等.典型天气条件下哈尔滨城市森林不同林型对PM2.5的调控作用研究[J].安徽农业科学,2016,44(5):175-179.
[11]GARDNER M W,DORLING S R.Statistical surface ozone models:An improved methodology to account fornon-linear behaviour[J]. Atmospheric environment,2000,34(1):21-34.
[12] CHALOULAKOU A,SAISANA M,SPYRELLIS N.Comparative assessment of neural networks and regression models for forecasting summertime ozone in Athens[J].Science of the total environment,2003,313(1/2/3):1-13.
[13]MCKENDRY I G.Evaluation of artificial neural networks for fine particulate pollution(PM10and PM2.5)forecasting[J].Journal of the air&waste management association,2002,52(9):1096-1101.
[14]KOLEHMAINEN M,MARTIKAINEN H,RUUSKANEN J.Neural networks and periodic components used in air quality forecasting[J].Atmospheric environment,2001,35(5):815-825.
[15]PAPANASTASIOU D K,MELAS D,KIOUTSIOUKIS I.Development and assessment of application of neural network and multiple regression models in order to predict PM10levels in a medium-sized Mediterranean city[J].Water,air&soil pollution,2007,182:325-334.
[16]曹虹.基于BP神经网络的交通流量预测[D].西安:长安大学,2012:30.
[17]王敏,邹滨,郭宇,等.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.
[18]石灵芝,邓启红,路婵,等.基于BP人工神经网络的大气颗粒物PM10质量浓度预测[J].中南大学学报(自然科学版),2012,43(5):1969-1974.
[19]宰松梅,郭冬冬,韩启彪,等.基于人工神经网络理论的土壤水分预测研究[J].中国农学通报,2011,27(8):280-283.
[2]孙淑萍,古润泽,张晶.北京城区不同绿化覆盖率和绿地类型与空气中可吸入颗粒物(PM10)[J].中国园林,2004,20(3):77-79.
[3]YU Y,SCHLEICHER N,NORRA S,et al.Dynamics and origin of PM2.5during a three-year sampling period in Beijing,China[J].Journal of environmental monitoring,2011,13(2):334-346.
[4]吴海龙,余新晓,师忱,等.PM2.5特征及森林植被对其调控研究进展[J].中国水土保持科学,2012,10(6):116-122.
[5] BECKETT K P,FREER-SMITH P H,TAYLOR G.Particulate pollution capture by urban trees:Effect of species and windspeed[J].Global change biology,2000,6(8):995-1003.
[6] LANGNER M. Reduction of airborne particulates by urban green[J].Bundesamt für naturschutz,2008,179:129-137.
[7]CAVANAGH J A E,ZAWAR-REZA P,WILSON J G.Spatial attenuation of ambient particulate matter air pollution within an urbanised native forest patch[J].Urban forestry&urban greening,2009,8(1):21-30.
[8]郭二果,王成,郄光发,等.北京西山典型游憩林空气悬浮颗粒物季节变化规律[J].东北林业大学学报,2010,38(10):55-57.
[9]殷杉,蔡静萍,陈丽萍,等.交通绿化带植物配置对空气颗粒物的净化效益[J].生态学报,2007,27(11):4590-4595.
[10]汪永英,孙琪,李昭,等.典型天气条件下哈尔滨城市森林不同林型对PM2.5的调控作用研究[J].安徽农业科学,2016,44(5):175-179.
[11]GARDNER M W,DORLING S R.Statistical surface ozone models:An improved methodology to account fornon-linear behaviour[J]. Atmospheric environment,2000,34(1):21-34.
[12] CHALOULAKOU A,SAISANA M,SPYRELLIS N.Comparative assessment of neural networks and regression models for forecasting summertime ozone in Athens[J].Science of the total environment,2003,313(1/2/3):1-13.
[13]MCKENDRY I G.Evaluation of artificial neural networks for fine particulate pollution(PM10and PM2.5)forecasting[J].Journal of the air&waste management association,2002,52(9):1096-1101.
[14]KOLEHMAINEN M,MARTIKAINEN H,RUUSKANEN J.Neural networks and periodic components used in air quality forecasting[J].Atmospheric environment,2001,35(5):815-825.
[15]PAPANASTASIOU D K,MELAS D,KIOUTSIOUKIS I.Development and assessment of application of neural network and multiple regression models in order to predict PM10levels in a medium-sized Mediterranean city[J].Water,air&soil pollution,2007,182:325-334.
[16]曹虹.基于BP神经网络的交通流量预测[D].西安:长安大学,2012:30.
[17]王敏,邹滨,郭宇,等.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.
[18]石灵芝,邓启红,路婵,等.基于BP人工神经网络的大气颗粒物PM10质量浓度预测[J].中南大学学报(自然科学版),2012,43(5):1969-1974.
[19]宰松梅,郭冬冬,韩启彪,等.基于人工神经网络理论的土壤水分预测研究[J].中国农学通报,2011,27(8):280-283.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |