Research On Atmospheric Environment Prediction System Based On Data Mining

With the rapid development of society and economy,air quality has increasingly become a problem closely related to people's life.In order to master the state of air quality in real time,monitoring stations of atmospheric environment have been set up to monitor the concentration of atmospheric pollutants and the level of air quality.Trend prediction of air quality change is a hot research topic nowadays.In this thesis,the historical data of air quality monitoring points of Beijing environmental protection monitoring center(China)are utilized to study the change rule and development trend of air quality index(AQI)at various monitoring stations in Beijing,China.And a medium and long term air quality prediction model is designed and put into effect.Which combine the conventional time series model with the popular support vector regression technology.Moreover,for validating the validity of the model,an atmospheric environment prediction system based on Python/Django web framework is set up.The main contents and innovations of this thesis are as follows:1.Air quality data acquisition and preprocessing.The Python web crawler is compiled to obtain the air quality information of air quality monitoring stations in Beijing and meteorological information from China meteorological data network.Web crawler script using Python's requests model to simulate the browser to send request,and using the DOM parser to complete the analysis of the returned XML document,and storing the captured data into the MongDB which is a database based on distributed file storage.Meanwhile,for data preprocessing,the abnormal value is set to null,the missing values are interpolated by Lagrange interpolation.2.Medium and long term air quality prediction of time series model combine with SVR.Times series is regarded as a combination of linear autocorrelation and nonlinear residual part.The ARMA model is established and the BIC criterion is used to determine the order.The predicted value of the ARMA model is taken as the linear part of the prediction.Based on SVR method,the residual rolling prediction of ARMA fitting is introduced as the nonlinear residual part of time series.3.Design and implementation of air quality index change prediction system.This system uses Python/Django web framework,combines online data display with prediction module,and completes online query of air quality and prediction display of atmospheric environment.The result shows that the proposed method can well predict the change of air quality.The function of air quality index change prediction is perfect and easy to operate.It can steadily realize the data information management of monitoring points,online atmospheric environment data query and online air quality prediction data query.