Research On Air Quality Condition And Prediction In Taiyuan Based On EMD-Prophet Model

Based on the data of air pollutant concentration and air quality index（AQI）in Taiyuan from 2015 to the first half of 2020,this paper explores the temporal distribution characteristics and evolution rules of air quality in Taiyuan.At the same time,it forecasts the air quality index in Taiyuan to study the future changes of air quality in Taiyuan.In this paper,R,Python and MATLAB were used to analyze the data.The main research contents and results are as follows:Firstly,the data were classified according to the time of year,season,and month,drew a descriptive statistical chart,directly observed at various air pollutants in the year,quarter,month,the change rule of analysis found that the air quality in Taiyuan city showed obvious seasonal characteristics,pollution exist different of pollutants during summer and winter seasons,also has obvious seasonal characteristics,especially the serious pollution in winter.Moreover,ozone production was particularly high in summer and other pollutants are high in winter.Then,according to the actual heating time,the heating period and the non-heating period were divided,and the multiple regression equation was established.According to the values of the regression coefficients,it can be concluded that PM₁₀ has a certain degree of influence on air quality in any period.The multiple regression model established in two time periods was used to predict the future AQI respectively,and the prediction effect was good.However,the unknown of independent variables made the model unable to predict the future situation,so it was necessary to choose a more appropriate model to predict the change of AQI.After that,in order to predict the future changes of air quality,a combined prediction model of empirical simulation decomposition（EMD）and Prophet model（EMD-Prophet model）is proposed.EMD is used to decompose the AQI value,and then the IMF components decomposed by EMD are used to construct the Prophet model and make predictions,and finally,the Prophet prediction results of each IMF component are integrated.Then,four models were established to predict the AQI value of Taiyuan.Finally,MAE,RMSE and MAPE values of EMD-Prophet model were compared with LSTM model,ARIMA（3,1,2）and Prophet model.MAE,RMSE and MAPE values of EMD-Prophet model were 31.7310,42.8702 and28.85%,which were all lower than the MAE,RMSE and MAPE of the other three models.The experimental results show that EMD-Prophet model can be better applied to predict AQI values.