时空图神经网络在PM2.5浓度预测中的应用综述

doi:10.11896/jsjkx.240700153

摘要/Abstract

摘要： 大气细颗粒物(PM_2.5)因对健康和环境有着严重的负面影响而备受公众关注。实现PM_2.5浓度的高精度时空预测,对于指导居民有效防范健康风险和协助环境监管部门制定科学的环境保护策略至关重要。研究旨在探索提高PM_2.5浓度预测准确性的有效方法,特别是时空图神经网络技术在该领域的应用潜力与挑战。首先,回顾了PM_2.5浓度预测方法的发展历程;随后,深入探讨了时空图神经网络与空气质量监测网络的结合点,包括图的构建策略;进一步,系统地概述了应用于PM_2.5浓度预测的时空图神经网络模型,并分析了在预测任务中应考虑的主要因素以及时空模块的设计;最后,从多源数据融合、动态图建模、数据稀疏性问题以及模型评估标准的缺失等多个维度,全面探讨了基于时空图神经网络的PM_2.5浓度预测所面临的挑战,并展望了未来可能的发展方向。

关键词: PM_2.5浓度预测, 深度学习, 图神经网络, 时空相关性, 多源数据

Abstract: Atmospheric fine particulate matter(PM_2.5) has garnered significant public attention due to its detrimental effects on health and the environment.Achieving high-precision spatial-temporal forecasting of PM_2.5 concentrations is crucial for guiding residents in effectively guarding against health risks and assisting environmental regulatory departments in formulating scientific environmental protection strategies.The study aims to explore effective methods for improving the accuracy of PM_2.5 concentration predictions,especially the application potential and challenges of spatial-temporal graph neural network technology in this field.It begins by reviewing the development history of PM_2.5 concentration forecasting methods,then delves into the integration of spatial-temporal graph neural networks with air quality monitoring networks,including strategies for constructing the graph.It further systematically outlines the spatial-temporal graph neural network models applied to PM_2.5 concentration prediction and analyzes the main factors to consider in forecasting tasks and the design of spatial-temporal modules.Finally,from multiple dimensions such as multi-source data fusion,dynamic graph modeling,issues of data sparsity,and the absence of model evaluation standards,it comprehensively discusses the challenges faced by PM_2.5 concentration prediction based on spatial-temporal graph neural networks and proposes possible directions for development.

Key words: PM_2.5 concentration forecasting, Deep learning, Graph neural network, Spatial-temporal correlation, Multisource data

中图分类号:

P228

唐博源, 李琦. 时空图神经网络在PM_2.5浓度预测中的应用综述[J]. 计算机科学, 2025, 52(8): 71-85. https://doi.org/10.11896/jsjkx.240700153

TANG Boyuan, LI Qi. Review on Application of Spatial-Temporal Graph Neural Network in PM_2.5 ConcentrationForecasting[J]. Computer Science, 2025, 52(8): 71-85. https://doi.org/10.11896/jsjkx.240700153

参考文献

[1]World Health Organization.WHO global air quality guidelines:particulate matter(PM_2.5 and PM₁₀),ozone,nitrogen dioxide,sulfur dioxide and carbon monoxide[M].World Health Organization,2021.
[2]NAN N,YAN Z P,ZHANG Y R,et al.Overview of PM_2.5 and health outcomes:focusing on components,sources,and pollutant mixture co-exposure[J].Chemosphere,2023,323:138181.
[3]WANG L L,LIU X J,LI D,et al.Geographical detection of spatial heterogeneity and drivers of PM_2.5 in the Yangtze River economic belt[J].Environmental Science,2022,43(3):1190-1200.
[4]ZHANG Y,BOCQUET M,MALLET V,et al.Real-time airquality forecasting,part I:History,techniques,and current status[J].Atmospheric Environment,2012,60:632-655.
[5]GAO Z Q,ZHOU X H.A review of the CAMx,CMAQ,WRF-Chem and NAQPMS models:Application,evaluation and uncertainty factors[J].Environmental Pollution,2024,343:123183.
[6]JEONG J I,PARK R J,YEH S W,et al.Statistical predictability of wintertime PM_2.5concentrations over East Asia using simple linear regression[J].Science of the Total Environment,2021,776:146059.
[7]CHEN J B,CHEN K Y,DING C,et al.An adaptive Kalman filtering approach to sensing and predicting air quality index values[J].IEEE Access,2020,8:4265-4272.
[8]SUN W,ZHANG H,PALAZOGLU A,et al.Prediction of 24-hour-average PM_2.5 concentrations using a hidden Markov mo-del with different emission distributions in Northern California[J].Science of the total environment,2013,443:93-103.
[9]ZHOU Y L,CHANG F J,CHANG L C,et al.Multi-output support vector machine for regional multi-step-ahead PM_2.5 forecasting[J].Science of the Total Environment,2019,651:230-240.
[10]FENG X,LI Q,ZHU Y J,et al.Artificial neural networks forecasting of PM_2.5 pollution using air mass trajectory based geographic model and wavelet transformation[J].Atmospheric Environment,2015,107:118-128.
[11]LIN Y C,LEE S J,OUYANG C S,et al.Air quality prediction by neuro-fuzzy modeling approach[J].Applied Soft Computing,2020,86:105898.
[12]HOU J X,LI Q,ZHU Y J,et al.Real-time forecasting system of PM_2.5 concentration based on spark framework and random fo-rest model[J].Science of Surveying and Mapping,2017,42(1):1-6.
[13]KARIMIAN H,LI Q,WU C L,et al.Evaluation of differentmachine learning approaches to forecasting PM_2.5 mass concentrations[J].Aerosol and Air Quality Research,2019,19(6):1400-1410.
[14]CHAE S,SHIN J,KWON S,et al.PM₁₀ and PM_2.5 real-time prediction models using an interpolated convolutional neural network[J].Scientific Reports,2021,11(1):11952.
[15]ONG B T,SUGIURA K,ZETTSU K.Dynamically pre-trained deep recurrent neural networks using environmental monitoring data for predicting PM_2.5[J].Neural Computing and Applications,2016,27:1553-1566.
[16]LI X,PENG L,YAO X J,et al.Long short-term memory neural network for air pollutant concentration predictions:Method development and evaluation[J].Environmental Pollution,2017,231:997-1004.
[17]HUANG G Y,LI X Y,ZHANG B,et al.PM_2.5 concentration forecasting at surface monitoring sites using GRU neural network based on empirical mode decomposition[J].Science of the Total Environment,2021,768:144516.
[18]DING C,WANG G Z,ZHANG X Y,et al.A hybrid CNN-LSTM model for predicting PM_2.5 in Beijing based on spatiotemporal correlation[J].Environmental and Ecological Statistics,2021,28(3):503-522.
[19]WU Z H,PAN S R,CHEN F W,et al.A comprehensive survey on graph neural networks[J].IEEE Transactions on Neural Networks and Learning Systems,2020,32(1):4-24.
[20]LI Y G,YU R,SHAHABI C,et al.Diffusion convolutional recurrent neural network:Data-driven traffic forecasting[J].ar-Xiv:1707.01926,2017.
[21]LIN Y J,MAGO N,GAO Y,et al.Exploiting spatiotemporalpatterns for accurate air quality forecasting using deep learning[C]//Proceedings of the 26th ACM SIGSPATIAL InternationalConference on Advances in Geographic Information Systems.2018:359-368.
[22]SCARSELLI F,GORI M,TSOI A C,et al.The graph neural network model[J].IEEE Transactions on Neural Networks,2009,20(1):61-80.
[23]BRUNA J,ZAREMBA W,SZLAM A,et al.Spectral networks and locally connected networks on graphs[J].arXiv:1312.6203,2013.
[24]GILMER J,SCHOENHOLZ S S,RILEY P F,et al.Neural mes-sage passing for quantum chemistry[C]//InternationalConfe-rence on Machine Learning.PMLR,2017:1263-1272.
[25]XU K Y L,HU W H,LESKOVEC J,et al.How powerful are graph neural networks?[J].arXiv:1810.00826,2018.
[26]LAM R,SANCHEZ-GONZALEZ A,WILLSON M,et al.Learning skillful medium-range global weather forecasting[J].Science,2023,382(6677):1416-1421.
[27]YU S,XIA F,LI S H,et al.Spatio-temporal graph learning for epidemic prediction[J].ACM Transactions on Intelligent Systems and Technology,2023,14(2):1-25.
[28]WANG C Y,LIN Z Y,YANG X C,et al.Hagen:Homophily-aware graph convolutional recurrent network for crime forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2022:4193-4200.
[29]VERDONE A,SCARDAPANE S,PANELLA M.Explainablespatio-temporal graph neural networks for multi-site photovoltaic energy production[J].Applied Energy,2024,353:122151.
[30]WANG S,LI Y R,ZHANG J,et al.PM_2.5-GNN:A domainknowledge enhanced graph neural network for PM_2.5 forecasting[C]//Proceedings of the 28th International Conference on Advances in Geographic Information Systems.2020:163-166.
[31]VELIČKOVIĆ P,CUCURULL G,CASANOVA A,et al.Graph attention networks[J].arXiv:1710.10903,2017.
[32]HUANG Y,YING J J C,TSENG V S.Spatio-attention embedded recurrent neural network for air quality prediction[J].Knowledge-Based Systems,2021,233:107416.
[33]BAI S,KOLTER J Z,KOLTUN V.An empirical evaluation of generic convolutional and recurrent networks for sequence mo-deling[J].arXiv:1803.01271,2018.
[34]DUN A,YANG Y N,LEI F.Dynamic graph convolution neural network based on spatial-temporal correlation for air quality prediction[J].Ecological Informatics,2022,70:101736.
[35]VASWANI A,SHAZEER N,PARMAR N,et al.Attention isall you need[C]//proceedings of the 31st International Confe-rence on Neural Information Processing Systems.2017:6000-6010.
[36]DEVLIN J,CHANG M W,LEE K,et al.Bert:Pre-training of deep bidirectional transformers for language understanding[J].arXiv:1810.04805,2018.
[37]ZHOU H Y,ZHANG S H,PENG J Q,et al.Informer:Beyond efficient transformer for long sequence time-series forecasting[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:11106-11115.
[38]LI P F,ZHANG T,JIN Y T.A spatio-temporal graph convolutional network for air quality prediction[J].Sustainability,2023,15(9):7624.
[39]SU L,GAO C C,REN X L,et al.Understanding the spatial representativeness of air quality monitoring network and its application to PM_2.5 in the mainland China[J].Geoscience Frontiers,2022,13(3):101370.
[40]QI Y L,LI Q,KARIMIAN H,et al.A hybrid model for spatiotemporal forecasting of PM_2.5 based on graph convolutional neural network and long short-term memory[J].Science of the Total Environment,2019,664:1-10.
[41]TENG M F,LI S W,XING J,et al.72-hour real-time forecasting of ambient PM_2.5 by hybrid graph deep neural network with aggregated neighborhood spatiotemporal information[J].Environment International,2023,176:107971.
[42]LIU X W,QIN M C,HE Y,et al.A new multi-data-driven spatiotemporal PM_2.5 forecasting model based on an ensemble graph reinforcement learning convolutional network[J].Atmospheric Pollution Research,2021,12(10):101197.
[43]GE L,WU K Y,ZENG Y,et al.Multi-scale spatiotemporalgraph convolution network for air quality prediction[J].Applied Intelligence,2021,51:3491-3505.
[44]ZHOU H Y,ZHANG F,DU Z H,et al.Forecasting PM_2.5 using hybrid graph convolution-based model considering dynamic wind-field to offer the benefit of spatial interpretability[J].Environmental Pollution,2021,273:116473.
[45]ZHOU H Y,ZHANG F,DU Z H,et al.A theory-guided graph networks based PM_2.5 forecasting method[J].Environmental Pollution,2022,293:118569.
[46]HETTIGE K H,JI J H,XIANG S L,et al.AirPhyNet:Har-nessing physics-guided neural networks for air quality prediction[J].arXiv:2402.03784,2024.
[47]LIAO H B,YUAN L,WU M,et al.Air quality prediction by integrating mechanism model and machine learning model[J].Science of the Total Environment,2023,899:165646.
[48]WANG C Y,ZHU Y M,ZANG T Z,et al.Modeling inter-station relationships with attentive temporal graph convolutional network for air quality prediction[C]//Proceedings of the 14th ACM International Conference on Web Search and Data Mi-ning.2021:616-634.
[49]CHEN L,XU J H,WU B Q,et al.Group-aware graph neural network for nationwide city air quality forecasting[J].ACM Transactions on Knowledge Discovery from Data,2023,18(3):1-20.
[50]SU I F,CHUNG Y C,LEE C,et al.Effective PM_2.5 concentration forecasting based on multiple spatial-temporal GNN for areas without monitoring stations[J].Expert Systems with Applications,2023,234:121074.
[51]OUYANG X,YANG Y,ZHANG Y,et al.Dual-channel spatial-temporal difference graph neural network for PM_2.5 forecasting[J].Neural Computing and Applications,2023,35(10):7475-7494.
[52]HAN J D,LIU H,XIONG H Y,et al.Semi-supervised air quality forecasting via self-supervised hierarchical graph neural network[J].IEEE Transactions on Knowledge and Data Enginee-ring,2023,35(5):5230-5243.
[53]NI Q J,WANG Y H,YUAN J Y.Adaptive scalable spatio-temporal graph convolutional network for PM_2.5 prediction[J].Engineering Applications of Artificial Intelligence,2023,126:107080.
[54]HAN J D,LIU H,ZHU H S,et al.Joint air quality and weather prediction based on multi-adversarial spatiotemporal networks[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:4081-4089.
[55]TAN J,LIU H,LI Y F,et al.A new ensemble spatio-temporal PM_2.5 prediction method based on graph attention recursive networks and reinforcement learning[J].Chaos,Solitons & Fractals,2022,162:112405.
[56]JIN X B,WANG Z Y,KONG J L,et al.Deep spatio-temporal graph network with self-optimization for air quality prediction[J].Entropy,2023,25(2):247.
[57]YU M Z,MASRUR A,BLASZCZAK-BOXE C.Predicting hourly PM_2.5concentrations in wildfire-prone areas using a spatiotemporal transformer model[J].Science of the Total Environment,2023,860:160446.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed