Based On Data Mining About Water Evaluation Prediction Research In Dongting Lake

The water pollution gradually become one of the important problems which affect the human life and development, and the research on the evaluation of forecasts of the water quality was particularly important.The paper was based on the14main monitoring cross section13indicators of water quality data of Dongting Lake in2006, and the monitoring data about the water quality of Total Nitrogen、Total Phosphorus、 COD of Dongting Lake in1997～2006, to evaluated of forecasts water quality of the Dongting Lake, to established the evaluation prediction model; Then gives the design scheme of early warning platform about the Dongting Lake water quality was based on ArcView GIS, the main work was as follows:(1) According to the single factor index method and comprehensive index method to evaluated the water quality, for the comprehensive pollution index of each section, among them, East Dongting Lake was the biggest cross section, the comprehensive pollution index was0.6432, which pollution was the most serious; the river mouth was the smallest section, which index was0.2355, the lightest pollution.(2) Based on the principal component analysis (PCA) to evaluate water quality, for the principal component scores and comprehensive scores of each section, among them, the section was east Dongting Lake, with the highest scores of1.3057,was the most polluted; and the score of the lowest section was river mouth, is0.8839, the lightest pollution.(3) By used the analytic hierarchy process(AHP) to evaluatde the water quality,the results show that the East Dongting Lake had the highest weighted score with0.4629, the most pollution; Pollution of river mouth section scored was the lowest, at0.2138, the lightest pullotion.(4) By used the grey correlation method to evaluate water quality, the East Dongting Lake got the highest weighted score with1.9731in each sections, the most polluted; Pollution of river mouth scored lowest, at0.4887, the lightest. And compared the characteristics and the effects of the evaluate methods of the above.(5) By determining white nization weight function and grey number, to established the water quality prediction model which based on the grey clustering method, and the historical data to test showed that the prediction accuracy was80%, and got the better effection.(6) According to the characters of Dongting lake water quality, the prediction error is small and the result is ideal.by established the hybrid forecastiong model of the BP neural network and the nonlinear regression analysis.(7) The results of prediction error is small to establish water quality prediction model of support vector machine (SVM) to predicted the water quality category. And take the comprehensive comparison of the characteristics and the effection about the three kinds of forecasting method.(8) Based on ArcView GIS, gives the early warning platform design of Dongting Lake water quality, which mainly included the database design, the function module design, the platform framework design and the module base design etc.