Haze Pollution Prediction Based On Ensemble Learning

In recent years,there have been frequent occurrences of fog and haze in many cities in my country,and the air quality has dropped significantly.At the same time,it has caused a lot of inconvenience to daily travel.The World Health Organization listed air pollution as one of the top ten threats to global health in 2019.At present,the Chinese government continues to increase its efforts to control environmental pollution,and improving smog pollution is one of the key directions in the process of environmental pollution control.Scholars at home and abroad have put forward many valuable research methods for the prediction of smog pollution,but there are still problems such as single index and poor fitting effect.In this thesis,the data sets of daily air pollution factors and meteorological factors in Xingtai City are selected as the research objects.First analyzed the smog pollution situation in Xingtai City in recent years,and found that the smog situation has been improving from 2014 to 2016,with slight fluctuations in 2017.The correlation analysis using the maximum information coefficient shows that the correlation between PM2.5 concentration and air pollution factor is strong,and the correlation strength with meteorological factors is weak overall.In the model construction stage,after reasonable preprocessing of the data set according to the changing characteristics of the time series data,the construction process of the long and short-term memory network,extreme gradient promotion and support vector regression model were explained respectively,and the method of parameter adjustment was mainly explained.Then using the ensemble learning(Stacking)method,the above three models are used as the base learner,and the extreme gradient lifting model is used as the meta-learner to fuse and make predictions,and a comparative model is set to judge the pros and cons of the fusion model.The results show that the fusion model constructed in this thesis performs well and has important reference significance for the prediction research of smog pollution in Xingtai City.