Polluted aquifer inverse problem solution using artificial neural networks

IF 0.8 Q4 WATER RESOURCES
M. Foddis, P. Ackerer, A. Montisci, G. Uras
{"title":"Polluted aquifer inverse problem solution using artificial neural networks","authors":"M. Foddis, P. Ackerer, A. Montisci, G. Uras","doi":"10.7343/as-2022-607","DOIUrl":null,"url":null,"abstract":"The problem of identifying an unknown pollution source in polluted aquifers, based on known contaminant concentrations measurement in the studied areas, is part of the broader group of issues, called inverse problems. This paper investigates the feasibility of using Artificial Neural Networks (ANNs) for solving the inverse problem of locating in time and space the source of a contamination event in a homogeneous and isotropic two dimensional domain. ANNs are trained in order to implement an input-output relationship which associates the position. Once the output of the system is known, the input is reconstructed by inverting the trained ANNs. The approach is applied for studying a theoretical test case where the inverse problem is solved on the basis of measurements of contaminant concentrations in monitoring wells located in the studied area. Groundwater pollution sources are characterized by varying spatial location and duration of activity. To identify these unknown pollution sources, concentration measurements data of monitoring wells are used. If concentration observations are missing over a length of time after an unknown source has become active, it is more difficult to correctly identify the unknown pollution source. In this work, a missing data scenario has been taken into consideration. In particular, a case where only one measurement has been made after the pollutant source interrupted its activity has been considered.","PeriodicalId":42515,"journal":{"name":"Acque Sotterranee-Italian Journal of Groundwater","volume":"8 1","pages":""},"PeriodicalIF":0.8000,"publicationDate":"2022-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"10","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Acque Sotterranee-Italian Journal of Groundwater","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.7343/as-2022-607","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"WATER RESOURCES","Score":null,"Total":0}
引用次数: 10

Abstract

The problem of identifying an unknown pollution source in polluted aquifers, based on known contaminant concentrations measurement in the studied areas, is part of the broader group of issues, called inverse problems. This paper investigates the feasibility of using Artificial Neural Networks (ANNs) for solving the inverse problem of locating in time and space the source of a contamination event in a homogeneous and isotropic two dimensional domain. ANNs are trained in order to implement an input-output relationship which associates the position. Once the output of the system is known, the input is reconstructed by inverting the trained ANNs. The approach is applied for studying a theoretical test case where the inverse problem is solved on the basis of measurements of contaminant concentrations in monitoring wells located in the studied area. Groundwater pollution sources are characterized by varying spatial location and duration of activity. To identify these unknown pollution sources, concentration measurements data of monitoring wells are used. If concentration observations are missing over a length of time after an unknown source has become active, it is more difficult to correctly identify the unknown pollution source. In this work, a missing data scenario has been taken into consideration. In particular, a case where only one measurement has been made after the pollutant source interrupted its activity has been considered.
污染含水层反问题的人工神经网络求解
根据所研究地区已知的污染物浓度测量,在受污染的含水层中识别未知污染源的问题,是被称为逆问题的更广泛问题的一部分。本文研究了利用人工神经网络(ann)在均匀、各向同性的二维区域中求解污染事件源在时间和空间上定位的逆问题的可行性。训练人工神经网络是为了实现与位置相关的输入输出关系。一旦系统的输出是已知的,输入是通过反向训练的人工神经网络重建。应用该方法研究了一个理论测试案例,在该案例中,根据研究区监测井的污染物浓度测量结果求解了反问题。地下水污染源具有不同的空间位置和活动时间特征。为了识别这些未知污染源,使用了监测井的浓度测量数据。如果在未知污染源变得活跃后的一段时间内缺少浓度观测,则更难以正确识别未知污染源。在这项工作中,考虑了丢失数据的情况。特别考虑了污染源中断其活动后只进行一次测量的情况。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
CiteScore
1.80
自引率
50.00%
发文量
36
审稿时长
8 weeks
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信