文章基本信息

标题：OUTDOOR PM2.5 CONCENTRATION PREDICTION METHOD BASED ON DEEPLEARNING UNDER SEVERE AIR POLLUTION WARNING
本地全文：下载
作者：Minglan Yuan ; Ying Long
期刊名称：Fresenius Environmental Bulletin
印刷版ISSN：1018-4619
出版年度：2021
卷号：30
期号：6
页码：7384-7394
语种：English
出版社：PSP Publishing
摘要：In order to improve the prediction accuracy ofthe PM2.5 concentration prediction method in a complex environment, an outdoor PM2.5 concentra- tion prediction method using deep learning under the early warning of heavy air pollution is proposed. Firstly, collect PM2.5 data from Tianjin monitoringbase stations, and perform correlation analysis and preprocessing with other air pollutants. Then, fully integrate the temporal memory characteristics of RNN and the advantages of CNN's automatic feature extraction to realize the extraction of PM2.5 concen- tration features. Finally, the sequence-to-sequence (Seq2Seq) model is used to predict the hourly and long-term concentrations of PM2.5 in the next 24 hours. Based on the MATLAB simulation platform, the experimental evaluation of the proposed method in terms of root mean square error (RMSE) and av- erage absolute error (MAE) shows that it can achieve rapid convergence and the prediction performance is relatively ideal. The RMSE and MAE are 16.24 re- spectively. And 13.05, both are better than other comparison methods and have certain application prospects.
关键词：Deep learning;PM25 concentration prediction;Prepro- cessing;Seq2Seq model;RNN;CNN