首页 | 官方网站   微博 | 高级检索  
     

基于树模型的北京市PM2.5预测效果对比分析
引用本文:李志生,梁锡冠,金宇凯,张华刚,欧耀春.基于树模型的北京市PM2.5预测效果对比分析[J].环境工程,2021,39(6):106-113.
作者姓名:李志生  梁锡冠  金宇凯  张华刚  欧耀春
作者单位:广东工业大学 土木与交通工程学院, 广州 510006
基金项目:广东省自然科学基金(S2011040003755);广东工业大学产学研合作项目"工业厂房室内甲醛与PM2.5治理项目"(18HK0031)
摘    要:在城市空气质量预测中,ρ(PM2.5)会受到气象条件和时间周期的影响。选取北京市全市为实验区域,对多种污染物浓度特征、时间特征及天气特征等进行分析,采用2019年33个空气质量监测站逐小时数据开展PM2.5预测实验,建立了基于特征的LightGBM (light gradient boosting machine) PM2.5质量浓度预测模型,分别与随机森林模型(RF)、梯度提升树模型(GBDT)、 XGBoost模型3个PM2.5浓度预测模型进行对比。结果表明:在PM2.5浓度预测精度方面,LightGBM模型最高,XGBoost模型次之,RF模型最差。LightGBM模型的PM2.5污染浓度预测准确率高于其他模型,R2为0.9614,且具有训练快、内存少等优点。LightGBM模型的5个评估指标均优于其他模型,说明其在PM2.5逐时预测上具有很好的稳定性和应用前景。

关 键 词:周期特征    机器学习    PM2.5影响因素    LightGBM    PM2.5预测
收稿时间:2020-07-22

A COMPARATIVE STUDY ON EDICTIVE EFFECT OF PM2.5 IN BEIJING BASED ON TREE MODELS
Affiliation:School of Civil and Transportation Engineering, Guangdong University of Technology, Guangzhou 510006, China
Abstract:In urban air quality forecast, the mass concentrations of PM2.5 were influenced by the meteorological conditions and time period. This article selected Beijing as the experimental area, analysing a variety of pollutants concentration characteristics, time characteristics and weather characteristics. The data by hour of 33 air quality monitoring stations in 2019 were used to carry out the PM2.5 forecast experiments, based on characteristics of LightGBM(light gradient boosting machine) PM2.5 mass concentration prediction model. The results showed that compared with random forests model(RF), gradient boosting decision tree model(GBDT), XGBoost model, LightGBM model had the highest prediction accuracy of PM2.5 concentration, XGBoost model came next, random forest model was the lowest. The accuracy of LightGBM model PM2.5 prediction was higher than other models, R2 was 0.9614, and training LightGBM model was fast and RAM needed less. LightGBM model on the five indicators were better than the rest of the model, and LightGBM model on PM2.5 hourly prediction had better stability and application prospects.
Keywords:
点击此处可从《环境工程》浏览原始摘要信息
点击此处可从《环境工程》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司    京ICP备09084417号-23

京公网安备 11010802026262号