首页 | 本学科首页   官方微博 | 高级检索  
     检索      

水库中重金属迁移的神经网络模拟
引用本文:张明,沈永明,甄宏,冯宇,尚宏志.水库中重金属迁移的神经网络模拟[J].环境科学与技术,2012,35(1):63-70.
作者姓名:张明  沈永明  甄宏  冯宇  尚宏志
作者单位:1. 大连理工大学海岸和近海工程国家重点实验室,辽宁大连,116023
2. 辽宁省环境科学研究院,辽宁沈阳,110031
基金项目:国家水体污染控制与治理科技重大专项子课题,国家自然科学基金
摘    要:重金属在水环境中的迁移转化是一个多因素影响的复杂过程,快速而准确地模拟水体中重金属的迁移一直是水环境研究领域的焦点。神经网络模型计算迅速、能以较高精度拟合任意非线性函数,在水文水质模拟中已有很多成功的应用,但对重金属的模拟还很少见到。文章以大伙房水库为例,用神经网络模型模拟了重金属镉、铜、汞和锌在水库中的浓度迁移过程。建模时依据水环境中重金属迁移转化机理并借鉴数学建模中边值问题的解决思想,确定了模型的输入和输出因子;对实际监测中重金属浓度低于检测方法的最低检出限的情况,将其浓度以0计算。模型对水库中镉和铜的浓度的模拟值与实测值的最大相对误差分别为17.5%和17.9%,对汞和锌的浓度的模拟值与实测值的确定系数分别为0.741和0.762,而对镉和铜的确定系数更是达到了0.96以上。对各重金属的模拟结果表明用神经网络模拟水环境中的重金属迁移是可行的,能快速得到比较准确的结果。文章建立的模拟大伙房水库重金属浓度迁移的神经网络模型对该水库水环境管理有一定的参考价值。

关 键 词:重金属浓度  神经网络  大伙房水库  边值问题  检出限

A Neural Network Based Model for Simulating the Transport of Heavy Metals in Reservoir
ZHANG Ming , SHEN Yong-ming , ZHEN Hong , FENG Yu , SHANG Hong-zhi.A Neural Network Based Model for Simulating the Transport of Heavy Metals in Reservoir[J].Environmental Science and Technology,2012,35(1):63-70.
Authors:ZHANG Ming  SHEN Yong-ming  ZHEN Hong  FENG Yu  SHANG Hong-zhi
Institution:1.State Key Laboratory of Coastal and Offshore Engineering,Dalian University of Technology,Dalian 116023,China; 2.Liaoning Academy of Environmental Science,Shenyang 110031,China)
Abstract:The transport of heavy metals in aquatic environment is complicated and can be affected by many factors.It is a focal point to simulate the heavy metal transport in water body quickly and precisely.Artificial neural networks(NN),which computes rapidly and can be implemented to any desired degree of accuracy in fitting any nonlinear function,have been successfully applied in modeling most of water quality indices,but not so when turn to heavy metals.The paper described an application of artificial neural network for modeling the transport of heavy metals such as Cd,Cu,Hg and Zn in Dahuofang Reservoir.According to the mechanism of heavy metal transport and using the idea of settling boundary value problems in mathematical model for reference,the input and output factors of the NN were determined.The heavy metal concentrations that can not be detected or are below the detection limit of the given detection method were computed as 0.The outputs of NN models for all four heavy metals were compared with the measured data.The maximum relative error(MaxRE) of Cd and Cu for each of their test datasets is 17.5% and 17.9% respectively.The coefficient of determination(R2) of Hg and Zn for their test datasets is 0.741 and 0.762 respectively,and that of Cd and Cu is even higher than 0.96.The results of NN model of each heavy metal show that NN is able to simulate the transport of heavy metals in aquatic environment and can provide reliable results.NN models for simulating the transport of heavy metals in the Reservoir is instructive to water environment management in reservoir.
Keywords:heavy metal concentration  neural network  Dahuofang Reservoir  boundary value problem  detection limit
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号