摘要：In the recent years, air pollution is a very seriousproblem in China and elsewhere, and it is a factor thatsignificantly affects the quality of human health. Fineparticulate matter (PM2.5) is considered to be the culprit ofhaze weather. Therefore, research affects the quality of humanlife on PM2.5 forecasting has received increasing attention.Knowing this information in advance is very important toprotect humans from health problems. This paper proposes anew prediction method, using the predicted value of theWeather Research and Forecasting (WRF) model as input data,increasing the atmospheric inversion factor as an additionalinput factor and constructing a municipal atmospheric pollutantresponse model through a random forest algorithm. we use3-fold cross-validation (CV) to evaluate model performance. Theresult of the experiment shows, compared with the traditionalatmospheric simulation method, this method has practicalapplication significance. The simulation results have improvedtimeliness and accuracy. It provides a simple and effectivemethod for PM2.5 prediction.
关键词：air pollution prediction (forecasting); PM2.5; random forest; weather research and forecasting (WRF) model
