| 720 | 0 | 92 |
| 下载次数 | 被引频次 | 阅读次数 |
PM2.5浓度的预测不仅有助于公众健康风险的预警，也为政府制定科学的治理政策提供重要依据。针对福建地区PM2.5浓度预测的研究相对较少，基于2014—2023年福建省PM2.5浓度观测数据，构建融合CNN与LSTM的混合深度学习模型，开展PM2.5浓度的短期预测研究。结果表明：1)模型在验证集上表现良好，R2为0.896 7,MAE为1.71μg·m-3,RMSE为2.40μg·m-3,具有较高的预测精度；2)预测值与实测值拟合关系为y=0.89x+2.27,体现出良好的趋势一致性；3)所建模型能有效捕捉时间序列特征，具备一定的中短期应用潜力；4)引入PM10等污染物特征可显著提升PM2.5预测精度，揭示多源污染因子协同影响的关键机制。
Abstract：The prediction of PM2.5 concentration not only helps to warn of public health risks, but also provides an important basis for the government to formulate scientific governance policies. There are relatively few studies on the prediction of PM2.5 concentration in Fujian Province. Based on the PM2.5 concentration observation data of Fujian Province from 2014 to 2023,this study constructed a hybrid deep learning model integrating CNN and LSTM to conduct short-term prediction of PM2.5 concentration. The results showed that, 1) the model performed well on the validation set, with an R2 of 0.896 7,a MAE of 1.71 μg·m-3,and a RMSE of 2.40 μg·m-3,showing high predictive accuracy. 2) The fitting relationship between the predicted value and the measured value was y=0.89x+2.27,which showed good trend consistency. 3) The constructed model could effectively capture the characteristics of time series and had certain medium-and short-term application potential. 4) Incorporating additional pollutant characteristics such as PM10 significantly improved the prediction accuracy of PM2.5 and revealed the key mechanism of the synergistic influence of multi-source pollution factors.
[1] ZHAO C X,LIN Z J,YANG L F,et al.A study on the impact of meteorological and emission factors on PM2.5 concentrations based on machine learning[J].Journal of Environmental Management,2025,376:124347.
[2] 国务院.大气污染防治行动计划(国发[2013]37号)[EB/OL].(2013-09-10)[2024-12-02].https://www.gov.cn/gongbao/content/2013/content_2496394.htm.[State Council.Air Pollution Prevention and Control Action Plan(Guofa[2013]No.37)[EB/OL].(2013-09-10)[2024-12-02].https://www.gov.cn/gongbao/content/2013/content_2496394.htm.]
[3] 国务院.打赢蓝天保卫战三年行动计划(国发[2018]22号)[EB/OL].(2018-06-27)[2024-12-02].https://www.gov.cn/gongbao/content/2018/content_5306820.htm.[State Council.Three-year action plan to win the battle to protect blue skies(Guofa[2018]No.22)[EB/OL] .(2018-06-27)[2024-12-02].https://www.gov.cn/gongbao/content/2018/content_5306820.htm.]
[4] 环境保护部.关于实施《环境空气质量标准》(GB3095-2012)的通知(环发[2012]11号)[EB/OL].(2012-02-29)[2024-12-02].https://www.gov.cn/zwgk/2012-03/02/content_2081004.htm.[Ministry of Environmental Protection.Notice on the implementation of the Ambient Air Quality Standard(GB3095-2012)(Huanfa[2012]No.11)[EB/OL].(2012-02-29)[2024-12-02].https://www.gov.cn/zwgk/2012-03/02/content_2081004.htm.]
[5] World Health Organization.Air quality guidelines global update 2005[EB/OL].(2006-08-26)[2024-12-02].https://www.who.int/publications/i/item/WHO-SDE-PHE-OEH-06.02.
[6] 杨复沫，贺克斌，马永亮，等.北京大气细粒子PM2.5的化学组成[J].清华大学学报(自然科学版),2002,42(12):1605-1608.[YANG F M,HE K B,MA Y L,et al.Chemical composition of PM2.5 fine particles in Beijing[J].Journal of Tsinghua University(Science and Technology),2002,42(12):1605-1608.]
[7] YU Y,DING F,MU Y,et al.High time-resolved PM2.5 composition and sources at an urban site in Yangtze River delta,China after the implementation of the APPCAP[J].Chemosphere,2020,261:127746.
[8] XIE M,LU X,DING F,et al.Evaluating the influence of constant source profile presumption on PMF analysis of PM2.5 by comparing long-and short-term hourly observation-based modeling[J].Environmental Pollution,2022,314,120273.
[9] 郭康.关于PM2.5影响因素的统计分析[D].秦皇岛：燕山大学，2014:4-18.[GUO K.Statistical analysis on influencing factors of PM2.5[D].Qinhuangdao:Yanshan University,2014:4-18.]
[10] 张艺耀，苗冠鸿，闫剑诗，等.影响PM2.5因素的多元统计分析与预测[J].资源节约与环保，2013(11):135-136.[ZHANG Y Y,MIAO G H,YAN J S,et al.Multivariate statistical analysis and prediction of factors influencing PM2.5[J].Resources Conservation and Environmental Protection,2013(11):135-136.]
[11] 陈潇，陈奕霖，甘晖，等.福建省县级碳排放时空特征和面板数据模型分析[J].福建师范大学学报(自然科学版),2023,39(5):83-92.[CHEN X,CHEN Y L,GAN H,et al.Spatiotemporal characteristics and panel data model analysis of county-level carbon emissions in Fujian Province[J].Journal of Fujian Normal University(Natural Science Edition),2023,39(5):83-92.]
[12] TAO J,ZHANG L,CAO J,et al.A review of current knowledge concerning PM2.5 chemical composition,aerosol optical properties and their relationships across China[J].Atmospheric Chemistry and Physics,2017,17(15):9485-9518.
[13] RIVA D R,MAGALHÃES C B,LOPES A A,et al.Low dose of fine particulate matter(PM2.5)can induce acute oxidative stress,inflammation and pulmonary impairment in healthy mice[J].Inhalation Toxicology,2011,23(5):257-267.
[14] DUAN Z.Effects of PM2.5 exposure on klebsiella pneumoniae clearance in the lungs of rats[J].China Medical Abstracts(Internal Medicine),2014,31(1):31-32.
[15] 彭斯俊，沈加超，朱雪.基于ARIMA模型的PM2.5预测[J].安全与环境工程，2014,21(6):125-128.[PENG S J,SHEN J C,ZHU X.PM2.5 prediction based on ARIMA model[J].Safety and Environmental Engineering,2014,21(6):125-128.]
[16] 陈军，高岩，张烨培，等.PM2.5扩散模型及预测研究[J].数学的实践与认识，2014,44(15):16-27.[CHEN J,GAO Y,ZHANG Y P,et al.Study on PM2.5 diffusion model and prediction[J].Practice and Understanding of Mathematics,2014,44(15):16-27.]
[17] ZHAO R,GU X X,XUE B,et al.Short period PM2.5 prediction based on multivariate linear regression model[J].PLOS One,2018,13(7):e0201011.
[18] 林雨，汪洋.基于土地利用回归模型的福建省PM2.5质量浓度时空分布研究[J].亚热带资源与环境学报，2020,15(4):29-39.[LIN Y,WANG Y.Spatiotemporal distribution of PM2.5 mass concentrations in Fujian Province based on land use regression model[J].Journal of Subtropical Resources and Environment,2020,15(4):29-39.]
[19] 周体鹏.基于克里金插值法的昆明市PM2.5预测[D].昆明：云南大学，2016:6-25.[ZHOU T P.PM2.5 prediction in Kunming based on Kriging interpolation method[D].Kunming:Yunnan University,2016:6-25.]
[20] 王敏，邹滨，郭宇，等.基于BP人工神经网络的城市PM2.5浓度空间预测[J].环境污染与防治，2013,35(9):63-66,70.[WANG M,ZOU B,GUO Y,et al.Spatial prediction of urban PM2.5 concentration based on BP artificial neural network[J].Environmental Pollution and Control,2013,35(9):63-66,70.]
[21] CAPILLA C.Prediction of hourly ozone concentrations with multiple regression and multilayer perceptron models[J].International Journal of Sustainable Development and Planning,2016,11(4):558-565.
[22] 张鸣敏.基于支持向量回归的PM2.5浓度预测研究[D].南京：南京信息工程大学，2015:1-16.[ZHANG M M.Research on PM2.5 concentration prediction based on support vector regression[D].Nanjing:Nanjing University of Information Science and Technology,2015:1-16.]
[23] WANG P,ZHANG H,QIN Z,et al.A novel hybrid-garch model based on ARIMA and SVM for PM2.5 concentrations forecasting[J].Atmospheric Pollution Research,2017,8(5):850-860.
[24] 韦晶，李占清.中国高分辨率高质量PM2.5数据集(2000—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/6168e75d-93ab-4e4a-b7ff-33152e49d0bf.[WEI J,LI Z Q.High-resolution and high-quality PM2.5 dataset in China(2000—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/6168e75d-93ab-4e4a-b7ff-33152e49d0bf.]
[25] Copernicus Climate Data Store.ERA5-Land hourly data from 1950 to present[DB/OL].(2018-06-14)[2025-03-06].https://cds.climate.copernicus.eu/datasets/reanalysis-era5-land?tab=overview.
[26] 韦晶，李占清.中国高分辨率高质量地面SO2数据集(2013—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/7630b0a2-58d7-4093-bd48-bbe69ddec7fd.[WEI J,LI Z Q.High-resolution and high-quality ground-level SO2 dataset in China(2013—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/7630b0a2-58d7-4093-bd48-bbe69ddec7fd.]
[27] 韦晶，李占清.中国高分辨率高质量地面NO2数据集(2008—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/cdd719d1-e0c0-49be-9f20-6a0ba54c8b38.[WEI J,LI Z Q.High-resolution and high-quality ground-level NO2 dataset in China(2008—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/cdd719d1-e0c0-49be-9f20-6a0ba54c8b38.]
[28] 韦晶，李占清.中国高分辨率高质量PM10数据集(2000—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/30b46d2f-78ee-4f3e-88ad-690383d47df5.[WEI J,LI Z Q.High-resolution and high-quality ground-level PM10 dataset in China(2000—2023)[DB/OL].(2023-02-13)[2025-03-06].https://data.tpdc.ac.cn/zh-hans/data/30b46d2f-78ee-4f3e-88ad-690383d47df5.]
[29] SPEARMAN C.The proof and measurement of association between two things[J].American Journal of Psychology,1904,15(1):72-101.
基本信息:
DOI：10.19687/j.cnki.1673-7105.2026.02.018
中图分类号:X513
引用信息:
[1]徐田田,甘晖.基于CNN-LSTM模型的福建省PM_(2.5)浓度预测[J].亚热带资源与环境学报,2026,21(02):183-190.DOI:10.19687/j.cnki.1673-7105.2026.02.018.
基金信息:
福建省软科学资助项目(2019R0051)
2026-03-16
2026-03-16
2026-03-16