Study On The Key Technologies Of The Basin Environmental Pollution Warning Based On Visual Data Mining

In recent years, The government and people pay more and more attention to the effect of environmental pollution, In today’s society, environmental pollution harm to society has become extremely hot issues; River Basin water pollution degree evaluation is not only to make administrative measures and the choice of governance technical premises, also is the establishment of early warning system to pollution and evaluation of effect on policy and technology. River pollution investigation results show that the valley along the river zone has formed the narrow pollution belt, especially the main city of Chongqing, Fuling, Wanzhou and Yichang coastal water quality has generally been pollution; Pollution information is very complex, now the most important is that how to use and capture environmental information, a comprehensive variety of contaminants, migration, transformation of data information is particularly important; Further analysis of data mining, identify the critical watershed warning information of each factor and pollution correlation and gets pollution early warning information general character and characteristic, dig out the hidden rules and patterns become a watershed research topic todayThis article main research work includes the following several aspects:(1) Introduction to artificial neural network and the basic theory of GA, the genetic algorithm and BP neural network advantages and limitations of characteristics, We propose a method based on multiple linear regression and BP neural network GA optimization algorithm, in order to make up for the weights and threshold selection on a random defect and improve the overall accuracy of fitting mathematical model.(2) According to the National Environmental Protection Bureau official website as well as statistical yearbook on real-time monitoring data, we collect and arrange the data which we obtained, Then, we construct the optimized network model of input and output variables, We establish a hybrid model which based on GA optimization multivariate linear regression and BP neural network. We use the Matlab language programming, simulation and prediction of river basin pollution level, contrast and analysis with the predictors of outcome of the traditional BP neural network, multiple linear regression method, testing mixed model accuracy and effectiveness.(3) We use GA-BP global optimization neural network model for water quality of Songhua River Basin pollution trend and prediction, testing prediction results with the monitoring data fit. Experimental results show the optimized network model has practical value.(4) We construct a basin early warning platform which based on Arc View GIS, then we construct the spatial and attribute database and combined platform and prediction model to be a optimized design, We have used a visual presentation of information mining from the data mining platform, We use GIS technique and river environment, realize information early warning to prevent and monitor the environment of river basin.