基于RFE-RF模型的太原市PM_(2.5)浓度预测研究

Study on PM_(2.5)Concentration Prediction Based on REF-RF Model

为了更好的利用高空气象要素采用机器学习的方法对太原市PM_(2.5)的浓度进行预测。采用太原市2015~2018年环境空气质量数据和NCEP再分析数据,将空气污染物PM_(2.5)浓度作为标签、根据RFE特征选择的结果,将最利于提升模型表现的预报因子作为输入,选用随机森林(RF)回归模型进行预测,同时构建3个对比模型以进一步验证RF模型的预测准确率。结果表明:RF模型的MAE、MAPE、RMSE分别为17.19、38.17%和26.0,与Lasso模型相比,分别降低了7.7%、5.1%和2.7%;相比于SVM预测模型的MAE、MAPE、RMSE分别降低了23.1%、15.3%和29.9%;相比于KNN预测模型,RF模型的MAE、MAPE、RMSE分别降低了17.2%、19.8%和15.2%。RF模型具有良好的预测效果,R^(2)达0.71,4种模型预测值与实测值的相关系数依次为0.76、0.78、0.82和0.84,RF模型的预报效果均好于Lasso模型、KNN模型和SVM模型。通过选取最优的RF预测模型应用到日常的环境空气质量预报业务中,将进一步提高太原市PM_(2.5)浓度预报的准确率,同时也为加强太原市的空气污染防治,实现环境综合管理和决策科学化提供了的重要科技手段。

In order to make better use of the high altitude meteorological elements,the machine learning method is used to predict the concentration of PM2.5 in Taiyuan City.In this paper,the ambient air quality data of Taiyuan from 2015 to 2018 and the NCEP reanalysis data were used,the PM_(2.5)concentration of air pollutants was used as the label,according to the RFE characteristics selecting result,the prediction factors that are most conducive to improving the performance of the model were used as the input,the random forest regression model was selected for prediction,and three comparative models were constructed to further verify the prediction accuracy of the RF model.The results showed that the MAE,MAPE and RMSE of RF model were 17.19,38.17 and 26.0,respectively,which were reduced by 7.7%,5.1%and 2.7%respectively compared with Lasso model.Compared with SVM prediction model,MAE,MAPE and RMSE decreased by 23.1%,15.3%and 29.9%respectively.Compared with KNN prediction model,the MAE,MAPE and RMSE of RF model decreased by 17.2%,19.8%and 15.2%respectively.The RF model has good prediction effect,with R^(2)reaching 0.71.The correlation coefficients between the predicted values of the four models and the measured values were 0.76,0.78,0.82 and 0.84 respectively.The prediction effect of the RF model is better than that of Lasso model,KNN model and SVM model.By selecting the best RF prediction model and applying it to the daily ambient air quality prediction business,this paper will further improve the accuracy of the PM_(2.5)concentration prediction in Taiyuan City,and also provide an important scientific and technological means for strengthening the prevention and control of air pollution in Taiyuan City,and realizing the comprehensive environmental management and scientific decision-making.