首页 | 本学科首页   官方微博 | 高级检索  
     

融合图卷积神经网络和注意力机制的PM2.5小时浓度多步预测
引用本文:傅颖颖,张丰,杜震洪,刘仁义. 融合图卷积神经网络和注意力机制的PM2.5小时浓度多步预测[J]. 浙江大学学报(理学版), 2021, 48(1): 74-83. DOI: 10.3785/j.issn.1008-9497.2021.01.011
作者姓名:傅颖颖  张丰  杜震洪  刘仁义
作者单位:1.浙江大学 浙江省资源与环境信息系统重点实验室,浙江 杭州 310028
2.浙江大学 地理信息科学研究所, 浙江 杭州 310027
基金项目:国家重点研发计划项目(2018YFB0505000);国家自然科学基金资助项目(41871287).
摘    要:PM2.5小时浓度多为单步预测。为实现PM2.5小时浓度的多步预测,基于“编码器-解码器”的序列-序列预测(Seq2Seq)模型,集合图卷积神经网络提取非欧式空间数据特征的能力以及注意力机制自适应关注特征的能力,提出了融合图卷积神经网络和注意力机制的PM2.5小时浓度多步预测(GCN_Attention_Seq2Seq)模型。并与Seq2Seq模型和使用了图卷积神经网络、未使用注意力机制的GCN_Seq2Seq模型进行了对照,以2015—2016年北京市22个空气质量监测站点的空气质量数据为样本进行实例验证,结果表明,Seq2Seq模型和图卷积神经网络(GCN)可对PM2.5小时浓度数据的时空依赖进行有效建模,注意力机制有助于减缓多步预测中的预测精度衰减,提升PM2.5小时浓度多步预测的精度。GCN_Attention_Seq2Seq模型可有效应用于多种长度的PM2.5浓度预测窗口。

关 键 词:图卷积  深度学习  注意力机制  PM2.5小时浓度多步预测  
收稿时间:2019-10-17

Multi-step prediction of PM2.5 hourly concentration by fusing graph convolution neural network and attention mechanism
FU Yingying,ZHANG Feng,DU Zhenhong,LIU Renyi. Multi-step prediction of PM2.5 hourly concentration by fusing graph convolution neural network and attention mechanism[J]. Journal of Zhejiang University(Sciences Edition), 2021, 48(1): 74-83. DOI: 10.3785/j.issn.1008-9497.2021.01.011
Authors:FU Yingying  ZHANG Feng  DU Zhenhong  LIU Renyi
Affiliation:1.Zhejiang Provincial Key Lab of GIS, Zhejiang University, Hangzhou 310028, China
2.Department of Geographic Information Science, Zhejiang University, Hangzhou 310027, China
Abstract:The current studies about PM2.5 hourly concentration prediction are mostly on single-step prediction.In order to achieve accurate prediction of PM2.5 hourly concentration at multiple moments in a single prediction task,this article proposes a multi-step prediction model of PM2.5 hourly concentration based on graph convolution neural network and attention mechanism,which is named GCN_Attention_Seq2Seq.The model based on Seq2Seq is able to extract the features of non-euclidean spatial data meantime pays attention to features adaptively.We take air quality data of 22 monitoring stations in Beijing from January 1st,2015 to December 29th,2016 as samples and compare GCN_Attention_Seq2Seq with GCN_Seq2Seq and Seq2Seq model. Results show that Seq2Seq and GCN can model spatio-temporal dependence effectively and the attention mechanism is helpful to improve the prediction accuracy and slow down the prediction accuracy decline in multi-step prediction,it indicates that the GCN_Attention_Seq2Seq model can be effectively applied to multi-step prediction of PM2.5 concentration.
Keywords:multi-step prediction of PM2.5 hourly concentration  graph convolution  attention mechanism  deep learning  
本文献已被 CNKI 等数据库收录!
点击此处可从《浙江大学学报(理学版)》浏览原始摘要信息
点击此处可从《浙江大学学报(理学版)》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