基于BP神经网络的重点行业企业周边土壤重金属污染预测及评价
|
摘要
对湖北省重点区域行业企业周边土壤理化指标和重金属含量进行监测;利用监测数据建立含有13输入、1个隐含层和6输出的3层BP神经网络模型,预测监测区域Mn、Co、V、Ag、Tl、Sb含量,综合重金属监测结果和预测结果,采用内梅罗指数对研究区域进行污染评价。结果表明,研究区域重金属存在不同程度超标情况,最大超标倍数范围为1.8～156.1倍;Mn、Co、V、Ag、Tl、Sb等6项重金属预测结果与实际测试结果相对误差范围在0.3%～19.9%,Mn、V、Ag、Tl、Sb在置信度为99%时均呈显著性相关(P<0.01,n=11),Co在置信度为95%时呈显著性相关(P<0.05,n=11),构建的BP神经网络预测模型具有良好的精准度;基于BP神经网络模型预测结果的内梅罗污染指数未超过警戒限的比例为77.3%,达轻度污染比例17.4%,达中度、重度污染比例均为4.0%。
The physical and chemical indicators and heavy metal content in the soil around the enterprises in key areas of Hubei Province were monitored. The monitoring data were used to establish a 3-layer BP neural network model with 13 inputs,1 hidden layer and 6 outputs.The content of Mn,Co,V,Ag,Tl,Sb in the monitoring area were predicted. The Nemerow index based on the monitoring and prediction results of heavy metals was used to evaluate the pollution of the area studied. The results showed that there were different levels of exceeding standard of heavy metals in the area studied. The maximum over-standard range was 1.8-156.1 times. The relative error between the prediction results of six heavy metals including Mn,Co,V,Ag,Tl and Sb and the actually tested results was ranged from 0.3% to 19.9%. Mn,V,Ag,Tl,and Sb were significantly correlated with the confidence of 99%(P<0.01,n=11).Co was significantly correlated with confidence of 95%(P<0.05,n=11). The BP neural network prediction model constructed had good accuracy. Based on the BP neural network model,the Nemerow pollution index not exceeding the warning limit took over a proportion of 77.3%,with the proportion of light pollution of 17.4% and the ratio of moderate to severe pollution of 4.0% each.
The physical and chemical indicators and heavy metal content in the soil around the enterprises in key areas of Hubei Province were monitored. The monitoring data were used to establish a 3-layer BP neural network model with 13 inputs,1 hidden layer and 6 outputs.The content of Mn,Co,V,Ag,Tl,Sb in the monitoring area were predicted. The Nemerow index based on the monitoring and prediction results of heavy metals was used to evaluate the pollution of the area studied. The results showed that there were different levels of exceeding standard of heavy metals in the area studied. The maximum over-standard range was 1.8-156.1 times. The relative error between the prediction results of six heavy metals including Mn,Co,V,Ag,Tl and Sb and the actually tested results was ranged from 0.3% to 19.9%. Mn,V,Ag,Tl,and Sb were significantly correlated with the confidence of 99%(P<0.01,n=11).Co was significantly correlated with confidence of 95%(P<0.05,n=11). The BP neural network prediction model constructed had good accuracy. Based on the BP neural network model,the Nemerow pollution index not exceeding the warning limit took over a proportion of 77.3%,with the proportion of light pollution of 17.4% and the ratio of moderate to severe pollution of 4.0% each.
引文
[1] MA X J,LU Z H,CHENG J L.Ecological risk assessment of open coal mine area[J].Environmental monitoring and assessment,2008,147:471-481.
[2] 蔡立梅,马瑾,周永章,等.东莞市农业土壤重金属的空间分布特征及来源解析[J].环境科学,2008,29(12):3496-3502.
[3] YALCIN M G,IIHAN S.Multivariate analyses to determine the origin of potenially harmful heavy metals in beach and dune sediments from Kizkalesi Coast (Mersin),Turkey[J].Bulletin of environmental contamination and toxicology,2008,81(12):57-68.
[4] 郑剑锋,焦继东,孙力平.基于神经网络的城市内湖水华预警综合建模方法研究[J].中国环境科学,2017,37(5):1872-1878.
[5] 秦绪佳,彭洁,徐菲,等.基于RBF网络的城市垃圾产量预测及可视化[J].中国环境科学,2018,38(2):792-800.
[6] 张恒德,张庭玉,李涛,等.基于BP神经网络的污染物浓度多模式集成预报[J].中国环境科学,2018,38(4):1243-1256.
[7] 汤勇,洪琪,王巧华,等.基于血线纹理特征和GA-BP神经网络的鸡种蛋性别鉴定[J].华中农业大学学报,2018,37(6):130-135.
[8] 陈明.MATLAB神经网络原理与实例精解[M].北京:清华大学出版社,2013:156-158.
[9] 胡大伟,卞新民,李思米,等.基于GRNN 网络模型的土壤重金属空间分布的研究[J].土壤通报,2007,38(2):334-340.
[10] LI Y,LI C K,TAO J J,et al.Study on spatial distribution of soil heavy metals in Huizhou city based on BP-ANN modeling and GIS[J].Procedia enviromental sciences,2011,10:1953-1960.
[11] 张钰,王让会,李成.基于BP神经网络的表层土壤重金属分布模拟[J].安全与环境工程,2014,21(2):51-56.
[12] 施彦,韩力群,廉小亲.神经网络设计方法与实例分析[M].北京:北京邮电大学出版社,2009:23-25.
[13] JIA Z Y,ZHOU S L,SU Q L.Comparison study on the estimation of the spatial distribution of regional soil metal(loid)s pollution based on kriging interpolation and BP neural network[J].International journal of environmental research and public health,2018,15(1):34-48.
[14] 中国环境监测总站,南京市环境监测中心站.HJ/T 166―2004土壤环境监测技术规范[S].北京:中国环境科学出版社,2004.
[15] 王天魏,蔡崇法,李朝霞,等.道路边际土壤重金属分布格局的神经网络模拟:以现代黄河三角洲为例[J].生态学报,2009,29(6):3154-3162.
[16] 杨娟,王昌全,李冰,等.基于BP神经网络的城市边缘带土壤重金属污染预测:以成都平原土壤Cd为例[J].土壤学报,2007,44(3):430-436.
[17] KUMAR R R,JYOTI K A.Artificial neural network (ANN) design for Hg-Se interactions and their effect on reduction of Hg uptake by radish plant[J].Journal of radioanalytical and nuclear chemistry,2009,283:797-801.
[18] 胡大伟,卞新民,王书玉,等.基于BP模型的南通市农田土壤重金属空间分布研究[J].安全与环境学报,2007,7(1):91-95.
[19] REZA S K,BARUAH U,SINGH S K,et al.Geostatistical and multivariate analysis of soil heavy metal contamination near coal mining area,Northeastern India[J].Environmental earth sciences,2015,73(9):5425-5433.
[20] 苏保林,王建平,贾海峰,等.密云水库流域非点源模型系统[J].清华大学学报(自然科学版),2006,46(3):351-359.
[21] 王牧童,王英伟,胡艳,等.基于BP模型的土壤重金属对其生物毒性的贡献分析[J].安全与环境学报,2016,16(4):348-352.
[22] 戴彬,吕建树,战金成,等.山东省典型工业城市土壤重金属来源、空间分布及潜在生态风险评价[J].环境科学,2015,36(2):507-515.
[23] 雷国建,刘千钧,陈志良,等.不同行业污染土壤重金属污染特征比较研究[J].土壤,2013,45(6):1023-1027.
[24] 范俊楠,贺小敏,陆泗进,等.重点行业企业周边土壤重金属污染现状及潜在生态危害评价[J].华中农业大学学报,2018,37(5):7-13.
[2] 蔡立梅,马瑾,周永章,等.东莞市农业土壤重金属的空间分布特征及来源解析[J].环境科学,2008,29(12):3496-3502.
[3] YALCIN M G,IIHAN S.Multivariate analyses to determine the origin of potenially harmful heavy metals in beach and dune sediments from Kizkalesi Coast (Mersin),Turkey[J].Bulletin of environmental contamination and toxicology,2008,81(12):57-68.
[4] 郑剑锋,焦继东,孙力平.基于神经网络的城市内湖水华预警综合建模方法研究[J].中国环境科学,2017,37(5):1872-1878.
[5] 秦绪佳,彭洁,徐菲,等.基于RBF网络的城市垃圾产量预测及可视化[J].中国环境科学,2018,38(2):792-800.
[6] 张恒德,张庭玉,李涛,等.基于BP神经网络的污染物浓度多模式集成预报[J].中国环境科学,2018,38(4):1243-1256.
[7] 汤勇,洪琪,王巧华,等.基于血线纹理特征和GA-BP神经网络的鸡种蛋性别鉴定[J].华中农业大学学报,2018,37(6):130-135.
[8] 陈明.MATLAB神经网络原理与实例精解[M].北京:清华大学出版社,2013:156-158.
[9] 胡大伟,卞新民,李思米,等.基于GRNN 网络模型的土壤重金属空间分布的研究[J].土壤通报,2007,38(2):334-340.
[10] LI Y,LI C K,TAO J J,et al.Study on spatial distribution of soil heavy metals in Huizhou city based on BP-ANN modeling and GIS[J].Procedia enviromental sciences,2011,10:1953-1960.
[11] 张钰,王让会,李成.基于BP神经网络的表层土壤重金属分布模拟[J].安全与环境工程,2014,21(2):51-56.
[12] 施彦,韩力群,廉小亲.神经网络设计方法与实例分析[M].北京:北京邮电大学出版社,2009:23-25.
[13] JIA Z Y,ZHOU S L,SU Q L.Comparison study on the estimation of the spatial distribution of regional soil metal(loid)s pollution based on kriging interpolation and BP neural network[J].International journal of environmental research and public health,2018,15(1):34-48.
[14] 中国环境监测总站,南京市环境监测中心站.HJ/T 166―2004土壤环境监测技术规范[S].北京:中国环境科学出版社,2004.
[15] 王天魏,蔡崇法,李朝霞,等.道路边际土壤重金属分布格局的神经网络模拟:以现代黄河三角洲为例[J].生态学报,2009,29(6):3154-3162.
[16] 杨娟,王昌全,李冰,等.基于BP神经网络的城市边缘带土壤重金属污染预测:以成都平原土壤Cd为例[J].土壤学报,2007,44(3):430-436.
[17] KUMAR R R,JYOTI K A.Artificial neural network (ANN) design for Hg-Se interactions and their effect on reduction of Hg uptake by radish plant[J].Journal of radioanalytical and nuclear chemistry,2009,283:797-801.
[18] 胡大伟,卞新民,王书玉,等.基于BP模型的南通市农田土壤重金属空间分布研究[J].安全与环境学报,2007,7(1):91-95.
[19] REZA S K,BARUAH U,SINGH S K,et al.Geostatistical and multivariate analysis of soil heavy metal contamination near coal mining area,Northeastern India[J].Environmental earth sciences,2015,73(9):5425-5433.
[20] 苏保林,王建平,贾海峰,等.密云水库流域非点源模型系统[J].清华大学学报(自然科学版),2006,46(3):351-359.
[21] 王牧童,王英伟,胡艳,等.基于BP模型的土壤重金属对其生物毒性的贡献分析[J].安全与环境学报,2016,16(4):348-352.
[22] 戴彬,吕建树,战金成,等.山东省典型工业城市土壤重金属来源、空间分布及潜在生态风险评价[J].环境科学,2015,36(2):507-515.
[23] 雷国建,刘千钧,陈志良,等.不同行业污染土壤重金属污染特征比较研究[J].土壤,2013,45(6):1023-1027.
[24] 范俊楠,贺小敏,陆泗进,等.重点行业企业周边土壤重金属污染现状及潜在生态危害评价[J].华中农业大学学报,2018,37(5):7-13.
© 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 |