Research Of Haze Forecasting Model Based On Support Vector Machine

In recent years,PM2.5 has gradually come into people's field of vision. At the same time,we realize that there is a more serious air quality problems. PM2.5 as a serious air pollution is closely related to our daily life. Air quality and its prediction methods,how to control and so on has become the focus of attention. Through the establishment of PM2.5 prediction model,we can master the change trend of PM2.5,and then we can analyze the changes of PM2.5,which has an important role in the research of PM2.5. The main research contents of this paper are as follows:1?Based on the existing research,we present a PM2.5 prediction method based on particle swarm optimization and genetic algorithm for parameter optimization of support vector machine(GA-SVR PSO-SVR model).2? In this paper,we use data to select the model and through the model to verify the data. We use the training the prediction model by meteorological data and air pollution data and select the prediction model by comparing the predicted results good results.3 ? Analysis the forecasting models in Baoding PM2.5 Case. By constructing the corresponding PSO-SVR and PM2.5 prediction model of GA-SVR,analysis and compare of the advantages and disadvantages of POS-SVR and GA-SVR prediction model,to verify the validity of the SVR model can be used to predict PM2.5.4? By using the factor analysis method of dimensionality reduction, select the data items with higher correlation on PM2.5,then through the KMO and sphericity test to verify whether the overall data can be used to predict,so as to reduce the amount of data,improve operational efficiency.