NEW APPROACH FOR ESTIMATION OF POLLUTANT LOAD BY USING ARTIFICIAL NEURAL NETWORK

摘要：

Pollutant load transported by rivers from non-point sources in watersheds is the major cause of eutrophication of closed water bodies such as lakes, reservoirs, and inner bays. Because it concentrates in a short time of flood event, regular water sampling cannot be adequate to monitor it. This study proposes a new type of monitoring technique of pollutant load in rivers: Optical characteristics of river water are monitored by a multi-item optical device. The relation between the sensor signals and the water qualities obtained from occasional sample analysis is modeled by Artificial Neural Network (ANN). After then, the time series of pollutant load can be produced from the optical signals. Field experiments were conducted in seven rivers flowing into Lake Kasumigaura. The ANN model trained by the data obtained in the year 2005 successfully produced the time series of pollutant load observed in the years 2006 and 2007. ANN models worked well in watersheds of different land use conditions if it was trained by the data obtained in each river.

关键词： artificial neural network pollutant load peculiar correlation optical sensor monitor

作者: Minghuan Liu Kenji Yoshimi Tadaharu Ishikawa Kentaro Kudo

作者单位: Dept.of Civil Eng., Tokyo Institute of Technology, Nagatsuta4259,Yokohama 226-8502, Japan Dept.of Environ.Sci.and Technoi., Tokyo Institute of Technology, Japan Dept.of Environ.Sci.and Technol., Tokyo Institute of Technology, Japan IDEA Consultants, Inc., Yokohama, Japan

会议类型: 国际会议

会议名称: 第16届亚太地区国际水利学大会暨第3届水工水力学国际研讨会(16th IAHR-APD Congress and 3rd Symoposium of IAHR-ISHS)

会议地点: 南京

会议语种:英文

页码: 683-688

在线出版日期: 2008-10-20（万方平台首次上网日期，不代表论文的发表时间）