Multi-step prediction of PM2.5 hourly concentration by fusing graph convolution neural network and attention mechanism
FU Yingying1,2, ZHANG Feng1,2, DU Zhenhong1,2, LIU Renyi1,2
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.
1 王平利，戴春雷，张成江.城市大气中颗粒物的研究现状及健康效应［J］.中国环境监测，2005，21（1）：85-89. DOI：10.3969/j.issn.1002-6002.2005.01.025 WANG P L，DAI C L，ZHANG C J.The study progress in the research for the particular in city air and its effect on human health［J］.Environmental Monitoring in China，2005，21（1）：85-89. DOI：10.3969/j.issn.1002-6002.2005.01.025 2 S?RENSEN M，DANESHVAR B，HANSEN M，et al. Personal PM2.5 exposure and markers of oxidative stress in blood［J］. Environmental Health Perspectives，2003，111（2）：161-165. DOI：10.1289/ehp.5646 3 ZHANG Y，BOCQUET M，MALLET V，et al. Real-time air quality forecasting，part I：History，techniques，and current status［J］.Atmospheric Environment，2012，60（32）：632-655. DOI：10.1016/j.atmosenv. 2012.06.031 4 徐文，黄泽纯，张倩宁.基于时空模型的PM2.5预测与插值［J］.江苏师范大学学报（自然科学版），2016，34（3）：70-75. DOI：10.3969/j.issn.2095-4298.2016. 03.016 XU W，HUANG Z C，ZHANG Q N.Prediction and interpolation of PM2.5 based on space-time model［J］.Journal of Jiangsu Normal University（Natural Science Edition），2016，34（3）：70-75. DOI：10.3969/j.issn.2095-4298.2016.03.016 5 范竣翔，李琦，朱亚杰，等.基于RNN的空气污染时空预测模型研究［J］.测绘科学，2017，42（7）：76-83. DOI：10.16251/j.cnki.1009-2307.2017.07.013 FAN J X，LI Q，ZHU Y J，et al.A Spatiotemporal prediction framework for air pollution based on deep RNN［J］.Science of Surveying and Mapping，2017，42（7）：76-83. DOI：10.16251/j.cnki.1009-2307. 2017.07.013 6 黄婕，张丰，杜震洪，等.基于RNN-CNN集成深度学习模型的PM2.5小时浓度预测［J］.浙江大学学报（理学版），2019，46（3）：370-379. DOI：10.3785/j.issn.1008-9497.2019.03.016 HUANG J，ZHANG F，DU Z H，et al.PM2.5 hourly concentration prediction based on RNN-CNN ensemble deep learning model［J］.Journal of Zhejiang University（Science Edition），2019，46（3）：370-379. DOI：10.3785/j.issn.1008-9497.2019.03.016 7 VASWANI A，SHAZEER N，PARMAR N，et al.Attention is all you need［C］//Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook：Curran Associoctes Inc. 2017：5998-6008. 8 CHEN J，ZHU Q，SHI D，et al.A multi-step wind speed prediction model for multiple sites leveraging spatio-temporal correlation［J］.Proceedings of the Chinese Society of Electrical Engineering，2019，39（7）：2093-2105. 9 GUO J，YU Y B，YANG C Y.Multi-step prediction of traffic load with all-attention mechanism［J］.Journal of Signal Processing，2019，35（5）：758-767. 10 GOYENA R.A computer movie simulating urban growth in the detroit region［J］.Journal of Chemical Information and Modeling，2019，53（9）：1689-1699. 11 GOODFELLOW I，BENGIO Y，COURVILLE A.Deep Learning［M］.Cambridge： The MIT Press，2016. 12 DEFFERRARD M，BRESSON X，VANDERGHEYNST P. Convolutional neural networks on graphs with fast localized spectral filtering［C］//Proceedings of the 30th International Conference on Neural Information Processing Systems. Barcelona：NIPS' 16，2016：3844-3852. 13 YU B，YIN H T，ZHU Z X.Spatio-temporal graph convolutional networks：A deep learning framework for traffic forecasting［C］//Proceedings of the 27th International Joint Conference on Artificial Intelligence. Stockholm：IJCAI，2018：3634-3640. DOI：10.24963/ijcai.2018/505 14 SIMONOVSKY M，KOMODAKIS N.Dynamic edge-conditioned filters in convolutional neural networks on graphs［C］//2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu：IEEE，2017. DOI：10.1109/cvpr.2017.11 15 KIPF T N，WELLING M.Semi-supervised classification with graph convolutional networks［C］//The 5th International Conference on Learning Representations. Toulon：ICLR，2017. 16 SUTSKEVER I，VINYALS O，LE Q V. Sequence to sequence learning with neural networks［C］//The 27th International Conference on Neural Information Processing Systems. Cambridge，MA：NIPS，2014：3104-3112. 17 CHO K，MERRIENBOER B V，GULCEHRE C，et al.Learning phrase representations using RNN encoder-decoder for statistical machine translation［C］.2014 Conference on Empirical Methods in Natural Language Processing. Doha：EMNLP，2014：1724-1734. DOI：10.3115/v1/d14-1179 18 RUMELHART D E，HINTON G E，WILLIAMS R J.Learning Internal Representations by Error Propagation［M］. Cambridge：MIT Press，1988：399-421. 19 HOCHREITER S，SCHMIDHUBER J.Long short-term memory［J］.Neural Computation，1997，9（8）：1735-1780. 20 WILLIAMS R J，ZIPSER D.A learning algorithm for continually running fully recurrent neural networks［J］.Neural Computation，1989，1（2）：270-280. DOI：10. 1162/neco.1989.1.2.270
2021, Vol. 48 
