Research On Atmospheric Environment Prediction Based On Data Mining Technology

With the deepending of China's environmental monitoring network coverage.A large number of monitoring data have been generated and accumulated by various environmental monitoring sites.However,these data are just used for data query,the potential value of the data has not been excavated.Therefore,it is necessary to use these historical data to find out the trend and rule of the change of atmospheric pollutant concentration,also it is very necessary to design and develop the atmospheric environmental prediction system.In this paper,the atmospheric environmental prediction model is analyzed and discussed in detail,and two kinds of atmospheric environmental prediction models are put forward,which are respectively for short-term(1 hours to 4 days)and long-term(7 days to 21 days)prediction cases.And on the basis of the two prediction models,the atmospheric environmental prediction system is set up,which is convenient for users to understand the future air pollutant concentration.The main research work of this paper is as follows:(1)Data acquisition and data preprocessing.Code the web crawler script and get the original data from Beijing environmental protection monitoring center,Web crawler script using HttpClient technology to simulate the browser to send requests and using Jsoup technology to complete the analysis of web page source code information.Using the method of data cleaning to the original data,delete the normal range and contradictory data and use the data transformation method to complete the different orders of magnitude,normalization of the dimension.(2)Research on short-term forecasting model based on multiple linear regression.The modeling method and the input factors are proposed to optimize the traditional multiple linear regression model,Experiments show that the modeling method is linearly regressive,and increasing the input factor seasonal factors and other pollutant concentrations can predict the future air pollutant concentration accurately and apply to the short-term prediction of atmospheric pollutant concentration.(3)Research on long-term forecasting model based on Genetic Neural Network.Aiming at the problem of insufficient global search capability of traditional BP neural network,easy to get into local optimum and slow training speed,this paper proposes the combination of BP neural network and genetic algorithm and changes the crossover probability and mutation probability of genetic algorithm with the change of adaptation degree.The improved neural network has the advantage of local searching for optimal solution,the global search and short training time,and it is suitable for the long-term forecast for the concentration of air pollutants.(4)Design and implementation of atmospheric environment prediction system.The system uses the MySQL database to store the data to meets the needs of algorithm model computation and system function realization.Using jQuery technology and Ajax technology to display the monitoring points on the map and display future air pollutant concentrations at the corresponding monitoring point,Using the Spring MVC framework to implement the background architecture design,the algorithm module is connected with the system function module to realize the function of the visual display of the browser and the calculation of the background environment forecasting model.