In view of the incompleteness of information utilization and the low prediction accuracy of PM2.5 prediction by multiple linear regression and time series models,a PM2.5 prediction method based on multiple time series (ARMAX) is proposed. This method introduces the delay order of PM2.5 influence factor on the time series in the regression term, extracts the residual error sequence information, and establishes the PM2.5 concentration prediction model. Firstly, pollutant data of Hefei in 2017 and 2018 were collected through the "Post-Weather Network";Next the data preprocessing and correlation analysis were completed;Then, the multiple linear regression model, time series model and ARMAX model of PM2.5 concentration prediction were respectively established;Finally, the prediction accuracy of the three models was compared by the three evaluation indicators of RMSE, MAE and Theil inequality coefficient. The result shows that the prediction accuracy of ARMAX model is better than that of single multiple linear regression model and time series model.
AO Xi-qin, ZHENG Yang, YU Yue-fen, WANG Jin-ting, LI Fan. PM2.5 Prediction Method Based on Multiple Time Series[J]. Journal of Chongqing Technology and Business University(Natural Science Edition）,2019,36(2):41-47