Data-driven Model Approach to Water Quality Analysis in a Reservoir |
---|
학술지명 International Symposium on advanced Intelligent Systems(ISIS2007)
저자 박상영,정남정,이요상
발표일 2007-09-05
|
In this study, the data-driven modelling approach for water quality forecasting in a reservoir was investigated. Three data-driven model algorithms such as model tree, ANN and RBF-NN implemented in WEKA machine learning software were employed for chlorophyll-a forecasting in the Youngdam reservoir, middle part of Korea. Model tree algorithm explicitly delineated dominant water quality parameters to build regression equation for chlorophyll-a forecasting. And those water quality parameters were used as an input in ANN and RBF-NN models. Model tree model showed good performance in terms of error between observed and predicted value, but in terms of correlation of the changing pattern of water quality RBF-NN model showed more good performance. The study result suggested that model tree algorithm could be used in input parameters selection for other models and the artificial intelligence model would efficiently be used in water quality forecasting in a reservoir that has non-linearity nature of water quality changing. |