The Study On Assessment Of Water Quality Of Nansi Lake Based On Bp Neural Network

Nansi Lake is an important transmission channel and storage lake in east route project of South-to-North Water Diversion, the water quality of the eastern water transferring line must meet theⅢclass standard of 'Environmental quality standard for surface water' according to the requirement of state council. In order to feel out the whole comprehensive water quality and spatial distribution in dry season before diversion, the study on evaluation of water quality of lake area and estuary in Nansi Lake before water diversion project has been done using improved BP neural network in this paper, and the evaluation results of comprehensive water quality in Nansi Lake is visualized by using surfer.Water environment is an open system far from balance, so there is plenty of nonlinear behavior. It is unreasonable and limited that the complexity and the nonlinear behaviors of water environment are not considered in traditional evaluation methods by used in our environmental assessment work at present. Neutral network method has been widely applied in comprehensive assessment of water quality, because it is a nonlinear system which has the ability of self-organization, self-learning, self-adaptive, fault-tolerant and anti-interference.Based on the investigation of the current situation of natural and society economy and pollution source in Nansi Lake drainage area, the water quality monitoring is implemented. BP neural network based on nonlinear academic and the application of mathematics software Matlab is introduced. According to the practical situation of the water pollution and monitoring factors in Nansi Lake, a BP network with a hiding level has been adopted. COD_Cr, NH₃-N, TP and TN are selected as the input nodes of the model. The output nodes of the model are water quality grades from class I to worse than classⅤ. Node numbers of hidden layer is determined as 30 by calculating repeatedly and the network structure of the water quality evaluation model is 4-30-6. Then the training samples are obtained according to Environmental quality standards for surface water (GB3838-2002). The model is trained using gradient descent algorithm by self-adapting gradient descent and momentum. Thus the comprehensive evaluation model is established. The water quality of the Up Lake and Down Lake and estuary is evaluated by using the above model, and the spatial distribution of water quality in Nansi Lake is depicted intuitively by using surfer. The results show that:1) The water quality of Up Lake seriously polluted is worse than Down Lake. The polluted zone with over standard is located at the most of Nanyang Lake, western area and centre of Dushan Lake, upper of Weishan Lake. On the whole, the most serious pollution area is located at the lower of the estuaries. Therefore, Point pollution brought by rivers of the lake is the main pollution factor.2) The percentage of the polluted estuaried with classⅣ, classⅤand worse than class Vis 52% in all estuaries which are monitored in Nansi Lake. The rivers mainly are Quangfu River, Xingfu River, Si River, Baima River, Chengguo River and Xuechengxiaosha River. The amount of pollutants and the over-standard instances are analysed. The pollutants over 90% in inflow rivers of Nansi Lake flow into Up Lake, the over standard rate and superstandard multiple of COD_Cr, TP, TN, NH₃-N and the above average in Up Lake's inflow rivers are greater than Down Lake's all mostly.In conclusion, the water quality of Up Lake is worse than Down Lake. The main pollution of the Nansi Lake comes from the rivers of Up Lake. Consequently, the key of the improvement of water quality in Nansi Lake is the comprehensive treatment about the rivers, especially the rivers of the Up Lake.