Water Pollution Prediction Of Huaihe River Based On Grey-BP Neural Network Model

Water is the primary souce of our human beings'life,which is the basic element of human beings'survival and development, as well as the most important factor of the nature ecology system.However,since the 1980s, as the social economy and the area along the Huaihe River are developing,water enviroment is becoming more and more deteriorated, which has become the key restrictive factor to the sustainable development of social economy.Accordingly, to use the water of Huaihe River reasonably and protect its sustainable ecology system is very influential for the social economy's sustainable development and mankind's survival .The condition of water quality has the direct influence on the social economy'sustainable development, and it is necessary to establish an effective forecast system to instruct the management of water pollution. PH value, dissolved oxygen, chemical oxygen demand and ammonia nitrogen are the important indexes of water quality. Basing on the research and the analysis of the four monitoring data of water quality from WangJia dam, Funan, Huaihe River, first of all, this thesis established the water quality's prediction model of Huaihe River in terms of Grey Model. Under the simulation of the real data, it shows the feasibility of this method with little sample data and its high accuracy, and its gradual worsening forecast effect as the time going on. Secondly, it establishes the water quality forecast model of Huaihe River based on BP Neural Network Model. Although its forecast effect is more accurate, it needs mountains of sample date and constant adjustment in the parameters to achieve high precision.Thirdly, considering the fact that Grey Model can weaken the volatility of data sequence and BP Neural Network's special nonlinear adaptive information processing capability, this thesis puts forward a new forecast model of Huaihe River's water quality in terms of Grey-BP Neural Network Model. It firstly predict the true value in terms of Grey Model. Then it treats the residual value between the true value and the prediction value as the input training sample of BP Neural Network model and selects proper BP neural network structure and the related parameters to train the sample. At last, it gain the final prediction value of Grey-BP Neural Network Model through plusing the residual value predicted by BP Neural Network Model and the prediction value of Grey Model on the condition of achieving the anticipate goal. The resarch indicates that the Grey-BP Neural Network Model requiring less sample data and having higher accurate precision can have a more accurate prediction of water condition in terms of less sample and information, and achieves comparative satisfied results. Through the contrast among the three forecast models and actual data, it is remarkably noticed that Grey-BP Neural Network Model which has better predictive effect and is more feasible than the other two models, has better applicative future and promotional value.