Application Research Of Artificial Neural Network Method On Water Quality Estimation In Yuan River Upriver

Yuan River comes from Yunwu Mountain in Duyun, Nanzhou, Which located in the east of Guizhou Province. The Yuan River belongs to the Dongting lake water system. In Hunan Province, the Yuan River has a length of 640 km, and in Huanhua, it's about 447 km. So the water condition has great meaning to the water environmental quality of Huaihua as well as Hunan province. With the increasing population, quicken city course, careless management of agriculture and industry and unreasonable resource development, the water quality is rapidly getting worse within the past few years. This paper used the upriver data monitored during the year 1994 to 2003 to estimate the water quality in that area, and discussed the application of Artificial Neural Network in water environmental quality. Water quality estimation neural network, lake and reservoir nutrition neural network and reservoir total phosphorous forecast neural network was established to estimate the water quality condition in the upriver Yuan River, the nutrition condition of Wuqiangxi reservoir and forecasted the total phosphorous concentration of the reservoir. Meanwhile, the single gene analytical method, principal component analytical method and seasonal Kendall check method were employed to make a synthetical estimation of the water quality in the drainage area.In the Artificial Neural Network establishing experiment, the influence of net structure, training function, the set of training error target, the structure of the samples. The experimental result showed that the performance of the ANN was effected by many factors, so abundant experiment should be taken to conferment an adaptive net. Because of the unsuitable resolving power of the water quality estimation result, the neutral network compositive method was used to achieve estimation. After composition, the resolving power has been raised dramatically, and the estimation result was more practical. As it proved that, the neutral network compositive method can efficiently elevate the extension ability of the ANN.Modified by the Artificial Neural Network, combined with other estimation method, the water quality condition of Yuan River is getting worse. Although a series of waste water treatment mature has been taken, it doesn't receive a remarkable effect. We should take effective action, to restrain the situation getting worse.