| Since the polluted smog appears in our country frequently,the environment problem has attracted many people's attention.Hence,how to control the damage and impact of the pollution is an urgent problem which needs to be solved.Note that it is a core to improving the quality of regional atmospheric environment by monitoring,forecasting and controlling the air quality index effectively.In this article,the statistics of the air quality index(in short AQI)of four cities are analyzed,which includes Beijing,Shanghai,Guangzhou and Lanzhou.In addition,the changed trends of air quality in next five days are predicted,which provides some practical applications to the air pollution controlling.Firstly,the statistics of a daily AQI of these cities are investigated for one year,and the research reveals that there are some large difference between different cities.More precisely,the degree of variability of AQI and the polluted weather at grade four and above are biggest in Beijing,while the air quality of Guangzhou is much more satisfied than other cities.In addition,we find that the air quality often changes with season and is much poor in winter.Furthermore,some suggestions in improving the air quality are given at last.In order to obtain some better results,we shall predict the real-time dates of AQI for these cities with different seasons.Firstly,the different sequences of statistics are pretreated by using EMD decomposition and normalized principle.Then,based on artificial intelligence algorithm to optimize the SVM model,we can forecast the seasonal AQI with multi-steps of these cities in five days.Finally,the prediction performance of models are compared by applying the standard evaluation of errors.Our results indicates that there does not exist an optimal and universal model in forecasting the seasonal AQI of different cities.Moreover,the prediction accuracy of the optimized SVM model can be approved,and one step prediction of the models are much better than that in multi-step prediction. |
PDF Full Text Request