一种卷积自编码深度学习的空气污染多站点联合预测模型

doi:10.12263/DZXB.20200258

电子学报 ›› 2022, Vol. 50 ›› Issue (6): 1410-1427.DOI: 10.12263/DZXB.20200258

所属专题：长摘要论文

一种卷积自编码深度学习的空气污染多站点联合预测模型

张波¹^,³, 陆云杰¹, 秦东明²^,⁴, 邹国建¹

^1.上海师范大学信息与机电工程学院，上海 200234
^2.同济大学电子与信息工程学院，上海 201804
^3.上海智能教育大数据工程技术研究中心，上海 200234
^4.中科三清科技有限公司，北京 100089

收稿日期:2020-03-14 修回日期:2021-07-25 出版日期:2022-06-25
- 作者简介:
- 张波男，1979年11月出生，江苏常州人.同济大学计算机应用技术专业工学博士.现为上海师范大学信息与机电工程学院院长、教授.主要研究方向为新一代人工智能与深度学习技术等.E-mail: zhangbo@shnu.edu.cn
  陆云杰男，1995年6月出生，江苏苏州人.上海师范大学信息与机电工程学院计算机应用技术工学硕士.主要研究方向为机器学习.E-mail: 1000459475@smail.shnu.edu.cn
  秦东明男，1981年4月出生，河南社旗人.同济大学电子与信息工程学院工学博士.主要研究方向为高性能计算、人工智能、深度学习等.
  邹国建男，1993年11月出生，安徽滁州人.上海师范大学信息与机电工程学院计算机应用技术专业工学硕士.主要研究方向为深度学习、知识图谱.
- 基金资助:
- 国家自然科学基金 (61802258); 上海自然科学基金 (18ZR1428300)

A Multi-Site Joint Air Pollution Prediction Model Based on Convolutional Auto-Encoder Deep Learning

ZHANG Bo¹^,³, LU Yun-jie¹, QIN Dong-ming²^,⁴, ZOU Guo-jian¹

^1.College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 200234, China
^2.College of Electronic and Information Engineering, Tongji University, Shanghai 201804, China
^3.Shanghai Engineering Research Center of Intelligent Education and Bigdata, Shanghai Normal University, Shanghai 200234, China
^4.3Clear Science & Technology Co. Ltd, Beijing 100193, China

Received:2020-03-14 Revised:2021-07-25 Online:2022-06-25 Published:2022-06-25
- Supported by:
- National Natural Science Foundation of China (61802258); Natural Science Foundation of Shanghai Municipality, China (18ZR1428300)

摘要/Abstract

摘要：

城市空气污染因空间扩散特性呈现出区域内的浓度高关联性.因此如何通过多个空气污染监测站的时空数据预测特定目标地点的污染情况，以解决站点分布不匀的问题，是一个重要的研究工作.本文结合空气污染物因素特性和气象因素的多维度影响，提出了一个利用区域内多站点空间监测数据实现特定目标站点的空气污染物浓度预测模型.该模型通过多层卷积神经网络（Convolutional Neural Network，CNN）实现城市多站点污染物浓度与气象数据之间的维度关联特征及空间关联特征学习，进而利用基于多层长短期记忆网络（Long Short-Term Memory，LSTM）的自编码网络实现多站点浓度的时序关联特征分析.实验通过真实数据集验证，所提出的预测模型获得了高于传统机器学习污染物浓度预测模型的预测准确度，且在多个城市数据集上验证了模型的泛化能力.

长摘要
随着空气污染问题日益严重，如何利用相关数据实现空气污染高效分析和预测，已成为实现空气污染高效治理所必须解决的迫切问题。城市空气污染因空间扩散特性呈现出区域内的浓度高关联性，而之前的研究大多聚焦于以单城市内的综合污染物数据来进行单城市的污染物浓度预测，并未考虑城市内多个分布不均匀站点之间的时空关系特征并进行联合预测。因此本文针对此瓶颈问题，提出了一个利用区域内多站点空间监测数据实现特定目标站点的空气污染物浓度预测模型CAELearning。该模型通过多层卷积神经网络（CNN）实现城市多站点污染物浓度与气象数据之间的维度关联特征及空间关联特征学习，进而利用基于多层长短期记忆网络（LSTM）的端到端模型架构网络实现多站点浓度的时序关联特征分析。模型同时考虑到了空间关联性即城市内多站点之间污染物浓度的相互影响，和时间关联性即污染物浓度前后时间段内的相互影响。通过真实数据集验证，该模型可以对未来一段时间内污染物浓度进行连续性预测，获得了高于传统机器学习污染物浓度预测模型的预测准确度，并在不同城市的预测上具有较好的泛化能力。同时，模型已经多次在国家级区域空气污染监测预报任务中作为实际辅助模型之一得到应用，体现出较好的价值。

关键词: 深度学习, 空气污染, 时空数据, 多站点联合预测, 卷积神经网络, 长短期记忆网络

Abstract:

Due to the nature of spatial diffusion, urban air pollution presents a high correlation with regional concentration feature. Therefore, how to use spatio-temporal related data from urban multiple air pollution detection sites to predict air pollution concentration of a special target location is an important research effort for solving the problem of uneven site distribution. Concerning the multi-dimensional impacts of the air pollutant factors’ features and the influence of meteorological factors, we propose an air pollutant concentration prediction model, which uses multi-site spatial detection data within a region to predict the concentration of the target station in this paper. This model is able to learn dimensional correlation characteristics and spatial correlation characteristics from multi-site pollutant concentration and meteorological data in the urban area through the multi-layer convolutional neural network, and then analyze the time-series correlation characteristics of multi-site concentration by utilizing the multi-layer auto-encoder network based long short-term memory network. The experimental results show that our proposed model obtains better performances than traditional machine learning based models under a real-world dataset, and meanwhile, the generalization performance of the proposed model has been examined based on multiple urban air pollution database.

Extended Abstract
As the air pollution problem becomes more and more serious, how to use relevant data to achieve efficient air pollution analysis and prediction has become an urgent problem that must be solved to achieve effective air pollution management. Urban air pollution exhibits high intra-regional concentration correlations due to spatial dispersion characteristics, while most previous studies focus on single-city pollutant concentration prediction with integrated pollutant data within a single city, without considering the characteristics of spatio-temporal relationships between multiple unevenly distributed stations within a city and making joint predictions. Therefore, this paper addresses this bottleneck problem and proposes a model called CAELearning (Learning net based on CNN and Auto-Encoder) for air pollutant concentration prediction at specific target sites using multi-site spatial monitoring data in the region by combining the characteristics of air pollutant factors and the multidimensional influence of meteorological factors. The proposed model considers the coupling relationship between the spatio-temporal characteristics of pollutants and meteorological data from the serial perspective of the model, realizes the dimensional association features and spatial association features learning between urban multi-site pollutant concentrations and meteorological data through a multilayer convolutional neural network (CNN), and then uses an end-to-end model architecture network based on a multilayer long short-term memory network (LSTM) to realize the temporal association features analysis of multi-site concentrations. The model takes into account both spatial correlation, i.e., the interaction of pollutant concentrations among multiple stations within a city, and temporal correlation, i.e., the interaction of pollutant concentrations in the time period before and after. Experiments based on real datasets indicate that the proposed prediction model can predict pollutant concentrations at specific sites in the next 24 hours based on pollutant concentrations at multiple sites in the past 72 hours as well as meteorological data, obtaining higher prediction accuracy than traditional machine learning pollutant concentration prediction models, and validating the generalization capability on multiple urban datasets. Meanwhile, the model has been applied as one of the practical auxiliary models in the national-level regional air pollution monitoring and forecasting tasks for several times, reflecting a better value.

Key words: deep learning, air pollution, spatio-temporal data, multiple sites joint prediction, convolutional neural network, long short-term memory network

中图分类号:

TP183

张波, 陆云杰, 秦东明, 等. 一种卷积自编码深度学习的空气污染多站点联合预测模型[J]. 电子学报, 2022, 50(6): 1410-1427.

Bo ZHANG, Yun-jie LU, Dong-ming QIN, et al. A Multi-Site Joint Air Pollution Prediction Model Based on Convolutional Auto-Encoder Deep Learning[J]. Acta Electronica Sinica, 2022, 50(6): 1410-1427.

图/表 22

图1 多站点的时空关系图

图2 输入模型的数据形式

图3 CAE-Learning预测模型框架

图4 自编码网络模型结构图（a）编码器部分的设计

图5 上海市多站点联合预测示意图

表1 数据集详细参数

参数名称	参数值
数据集时间跨度	2014.5.13—2018.3.24
训练集	2014.5.13—2017.12.31
测试集	2018.1.1—2018.3.24
预测未来时长	24 h
历史数据时长	72 h
矩阵大小（站点数*特征数）	［9，16］
最大迭代次数（Epoch）	100

图6 CAE-Learning的多站点联合预测结果

图7 CNN-LSTM的多站点联合预测结果

表2 每个模型的RMSE、MAE、相关系数和训练时间

模型	RMSE	MAE	相关系数	训练时间
BP	15.857	10.284	0.875	181 s
RNN	15.670	9.398	0.905	197 s
LSTM	9.688	7.139	0.958	304 s
CNN	27.068	21.048	0.980	871 s
CNN+LSTM	22.902	19.033	0.897	980 s
CNN-RNN	15.710	12.932	0.951	1219 s
CNN-LSTM	9.173	7.108	0.975	1342 s
CAE-Learning	8.880	7.001	0.980	2249 s

图8 CNN+LSTM并行的多站点联合预测结果

图9 CAE-Learning的多站点联合预测结果

图10 CNN-LSTM的多站点联合预测结果

表3 模型在杭州市数据集下的RMSE、MAE、相关系数和训练时间

模型	RMSE	MAE	相关系数	训练时间
CNN-LSTM	23.153	21.909	0.899	1576 s
CAE-Learning	18.249	16.535	0.945	2785 s

图11 CAE-Learning的多站点联合预测结果

图12 CNN-LSTM的多站点联合预测结果

表4 模型在苏州市数据集下的RMSE、MAE、相关系数和训练时间

模型	RMSE	MAE	相关系数	训练时间
CNN-LSTM	24.529	20.693	0.893	1657 s
CAE-Learning	18.249	16.535	0.945	2269 s

图13 CAE-Learning的多站点联合预测结果

图14 CNN-LSTM的多站点联合预测结果

表5 模型在重庆市数据集下的RMSE、MAE、相关系数和训练时间

模型	RMSE	MAE	相关系数	训练时间
CNN-LSTM	37.881	28.744	0.407	1867 s
CAE-Learning	18.735	15.662	0.923	2487 s

图15 CAE-Learning的多站点联合预测结果

图16 CNN-LSTM的多站点联合预测结果

表6 模型在北京市数据集下的RMSE、MAE、相关系数和训练时间

模型	RMSE	MAE	相关系数	训练时间
CNN-LSTM	36.158	24.504	0.601	1578 s
CAE-Learning	18.727	15.260	0.903	2278 s

参考文献 39

1	SCHRAUFNAGELD E, BALMESJ R, COWLC T, et al. Air pollution and noncommunicable diseases: A review by the Forum of International Respiratory Societies' Environmental Committee, Part 2: Air pollution and organ systems[J]. Chest, 2019, 155(2): 417-426.
2	ZOUB, YOUJ, LINY, et al. Air pollution intervention and life-saving effect in China[J]. Environment International, 2019, 125: 529-541.
3	CHANGX, WANGS, ZHAOB, et al. Contributions of inter-city and regional transport to PM_2.5 concentrations in the Beijing-Tianjin-Hebei region and its implications on regional joint air pollution control[J]. Science of the Total Environment, 2019, 660: 1191-1200.
4	LEONGW C, KELANIR O, AHMADZ. Prediction of air pollution index(API) using support vector machine(SVM)[J]. Journal of Environmental Chemical Engineering, 2020, 8(3): 103208.
5	MACIĄGP S, KASABOVN, KRYSZKIEWICZM, et al. Air pollution prediction with clustering-based ensemble of evolving spiking neural networks and a case study for London area[J]. Environmental Modelling & Software, 2019, 118: 262-280.
6	CABANEROSS M S, CALAUTITJ K, HUGHESB R. A review of artificial neural network models for ambient air pollution prediction[J]. Environmental Modelling & Software, 2019, 119: 285-304.
7	CORANIG, SCANAGATTAM. Air pollution prediction via multi-label classification[J]. Environmental Modelling & Software, 2016, 80: 259-264.
8	LIA, XUX. A new PM_2.5 air pollution forecasting model based on data mining and BP neural network model[C]//Proceedings of the 2018 3rd International Conference on Communications, Information Management and Network Security(CIMNS 2018). Berlin: Atlantis Press, 2018: doi:10.2991/cimns-18.2018.25.
9	FANGR, POUYANFARS, YANGY, et al. Computational health informatics in the big data age: A survey[J]. ACM Computing Surveys, 2016, 49(1): 1-36.
10	WAXA, ASENCIOR, CARTERD R. Thinking big about big data[J]. Industrial and Organizational Psychology, 2015, 8(4): 545-550.
11	秦喜文, 刘媛媛, 王新民, 等. 基于整体经验模态分解和支持向量回归的北京市PM2.5预测[J]. 吉林大学学报(地球科学版), 2016, 46(2): 563-568.
	QINX W, LIUY Y, WANGX M, et al. PM2.5 prediction of Beijing city based on ensemble empirical mode decomposition and support vector regression[J]. Journal of Jilin University(Earth Science Edition), 2016, 46(2): 563-568. (in Chinese)
12	桑海峰, 王传正, 吕应宇, 等. 基于多信息流动卷积神经网络的行人再识别[J]. 电子学报, 2019, 47(2): 351-357.
	SANGH F, WANGC Z, LVY Y, et al. Person re-identification based on multi-information flow convolutional neural network[J]. Acta Electronica Sinica, 2019, 47(2): 351-357. (in Chinese)
13	DEVLINJ, CHANGM W, LEEK, et al. Bert: Pre-training of deep bidirectional transformers for language understanding[EB/OL]. (2018)[2020]. .
14	LEEJ, YOONW, KIMS, et al. BioBERT: Pre-trained biomedical language representation model for biomedical text mining[EB/OL]. (2019)[2020]. .
15	廖勇, 花远肖, 姚海梅, 等. 高速移动环境下基于深度学习的信道估计方法[J]. 电子学报, 2019, 47(8): 1701-1707.
	LIAOY, HUAY X, YAOH M, et al. Channel estimation method based on deep learning in high-speed mobile environments[J]. Acta Electronica Sinica, 2019, 47(8): 1701-1707. (in Chinese)
16	GONGZ, ZHONGP, YUY, et al. A CNN with multiscale convolution and diversified metric for hyperspectral image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(6): 3599-3618.
17	刘启超, 肖亮, 刘芳, 等. SSCDenseNet: 一种空-谱卷积稠密网络的高光谱图像分类算法[J]. 电子学报, 2020, 48(4): 751-762.
	LIUQ C, XIAOL, LIUF, et al. SSCDenseNet: A spectral-spatial convolutional dense network for hyperspectral image classification[J]. Acta Electronica Sinica, 2020, 48(4): 751-762. (in Chinese)
18	SÜLOI, KESKINS R, DOGANG, et al. Energy efficient smart buildings: LSTM neural networks for time series prediction[C]//2019 International Conference on Deep Learning and Machine Learning in Emerging Applications(Deep-ML). New York: IEEE, 2019: 18-22.
19	ZHAOJ, DENGF, CAIY, et al. Long short-term memory-Fully connected(LSTM-FC) neural network for PM2.5 concentration prediction[J]. Chemosphere, 2019, 220: 486-492.
20	ZHAOP, ZETTSUK. Convolution recurrent neural networks based dynamic transboundary air pollution predictiona[C]//2019 IEEE 4th International Conference on Big Data Analytics(ICBDA). New York: IEEE, 2019: 410-413.
21	秦东明, 丁志军, 金玉鹏, 等. 基于自编码网络的空气污染物浓度预测[J]. 同济大学学报(自然科学版), 2019, 47(5): 681-687.
	QIND M, DINGZ J, JINY P, et al. An air pollutant prediction model based on auto-encoder network[J]. Journal of Tongji University(Natural Science), 2019, 47(5): 681-687. (in Chinese)
22	FRANCKU, ODEHS, WIEDENSOHLERA, et al. The effect of particle size on cardiovascular disorders—the smaller the worse[J]. Science of the Total Environment, 2011, 409(20): 4217-4221.
23	CHEMELC, FISHERB E A, KONGX, et al. Application of chemical transport model CMAQ to policy decisions regarding PM2.5 in the UK[J]. Atmospheric Environment, 2014, 82(1): 410-417.
24	WANGY, KONGT. Air quality predictive modeling based on an improved decision tree in a weather-smart grid[J]. IEEE Access, 2019, 7: 172892-172901.
25	GNEITINGT, RAFTERYA E, WESTVELDA H, et al. Calibrated probabilistic forecasting using ensemble model output statistics and minimum CRPS estimation[J]. Monthly Weather Review, 2010, 133(5): 1098-1118.
26	BALACHANDRANS, CHANGH H, PACHONJ E, et al. Bayesian-based ensemble source apportionment of PM2.5[J]. Environmental Science & Technology, 2013, 47(23): 13511-13518.
27	SARVIF, NADALIA, KHODADOSTM, et al. Application of poisson hidden Markov model to predict number of PM2.5 exceedance days in tehran during 2016-2017[J]. Avicenna Journal of Environmental Health Engineering, 2017, 4(1): 58031.
28	WANGM, YUANZ, ZHANGX, et al. Construction of air quality evaluation system based on FCM algorithm and BP neural network[J]. Agricultural Biotechnology, 2018, 7(5): 274-276.
29	CHENY. Prediction algorithm of PM2.5 mass concentration based on adaptive BP neural network[J]. Computing, 2018, 100(8): 825-838.
30	LECUNY, BENGIOY, HINTONG. Deep learning[J]. Nature, 2015, 521(7553): 436-444.
31	LEV D, BUIT C, CHAS K. Spatiotemporal deep learning model for citywide air pollution interpolation and prediction[C]//2020 IEEE International Conference on Big Data and Smart Computing(BigComp). Busan: IEEE, 2020: 19534862.
32	WANGJ, SONGG. A deep spatial-temporal ensemble model for air quality prediction[J]. Neurocomputing, 2018, 314: 198-206.
33	LINY, MAGON, GAOY, et al. Exploiting spatiotemporal patterns for accurate air quality forecasting using deep learning[C]//Proceedings of the 26th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems. New York: ACM, 2018: 359-368.
34	QIZ, WANGT, SONGG, et al. Deep air learning: Interpolation, prediction, and feature analysis of fine-grained air quality[J]. IEEE Transactions on Knowledge and Data Engineering, 2018, 30(12): 2285-2297.
35	WENC C, LIUS F, YAOX J, et al. A novel spatiotemporal convolutional long short-term neural network for air pollution prediction[J]. Science of The Total Environment, 2019, 654(1): 1091-1099.
36	ONGB T, SUGIURAK, ZETTSUK. Dynamically pre-trained deep recurrent neural networks using environmental monitoring data for predicting PM2.5[J]. Neural Computing and Applications, 2016, 27(6): 1553-1566.
37	HUANGC J, KUOP H. A deep CNN-LSTM model for particulate matter(PM2.5) forecasting in smart cities[J]. Sensors, 2018, 18(7): 2220.
38	KAREVANZ, SUYKENSJ A K. Transductive LSTM for time-series prediction: An application to weather forecasting[J]. Neural Networks, 2020, 125: 1-9.
39	常亮, 邓小明, 周明全, 等. 图像理解中的卷积神经网络[J]. 自动化学报, 2016, 42(9): 1300-1312.
	CHANGL, DENGX M, ZHOUM Q, et al. Convolutional neural networks in image understanding[J]. Acta Automatic Sinica, 2016, 42(9): 1300-1312. (in Chinese)

[1]	李钦, 刘伟, 牛朝阳, 宝音图, 惠周勃. 低信噪比下基于分裂EfficientNet网络的雷达信号调制方式识别[J]. 电子学报, 2023, 51(3): 675-686.
[2]	黄赟, 张帆, 郭威, 陈立, 羊光. 一种基于数据标准差的卷积神经网络量化方法[J]. 电子学报, 2023, 51(3): 639-647.
[3]	范兵兵, 何庭建, 张聪炫, 陈震, 黎明. 联合遮挡约束与残差补偿的特征金字塔光流计算方法[J]. 电子学报, 2023, 51(3): 648-657.
[4]	吕杭, 蒋明峰, 李杨, 张鞠成, 王志康. 基于混合时频域特征的卷积神经网络心律失常分类方法的研究[J]. 电子学报, 2023, 51(3): 701-711.
[5]	张聿远, 闫文君, 张立民. 基于多模态特征融合网络的空时分组码识别算法[J]. 电子学报, 2023, 51(2): 489-498.
[6]	许新征, 李杉. 基于特征膨胀卷积模块的轻量化技术研究[J]. 电子学报, 2023, 51(2): 355-364.
[7]	丁琪, 田萱, 孙国栋. 基于注意力增强的热点感知新闻推荐模型[J]. 电子学报, 2023, 51(1): 93-104.
[8]	李滔, 董秀成, 林宏伟. 基于深监督跨尺度注意力网络的深度图像超分辨率重建[J]. 电子学报, 2023, 51(1): 128-138.
[9]	郭晓轩, 冯其波, 冀振燕, 郑发家, 杨燕燕. 多线激光光条图像缺陷分割模型研究[J]. 电子学报, 2023, 51(1): 172-179.
[10]	贾童瑶, 卓力, 李嘉锋, 张菁. 基于深度学习的单幅图像去雾研究进展[J]. 电子学报, 2023, 51(1): 231-245.
[11]	何滢婕, 刘月峰, 边浩东, 郭威, 张小燕. 基于Informer的电池荷电状态估算及其稀疏优化方法[J]. 电子学报, 2023, 51(1): 50-56.
[12]	张永梅, 孙捷. 基于动静态特征双输入神经网络的咳嗽声诊断COVID-19算法[J]. 电子学报, 2023, 51(1): 202-212.
[13]	袁海英, 成君鹏, 曾智勇, 武延瑞. Mobile_BLNet：基于Big-Little Net的轻量级卷积神经网络优化设计[J]. 电子学报, 2023, 51(1): 180-191.
[14]	王神龙, 雍宇, 吴晨睿. 基于伪孪生神经网络的低纹理工业零件6D位姿估计[J]. 电子学报, 2023, 51(1): 192-201.
[15]	吴靖, 叶晓晶, 黄峰, 陈丽琼, 王志锋, 刘文犀. 基于深度学习的单帧图像超分辨率重建综述[J]. 电子学报, 2022, 50(9): 2265-2294.

一种卷积自编码深度学习的空气污染多站点联合预测模型

A Multi-Site Joint Air Pollution Prediction Model Based on Convolutional Auto-Encoder Deep Learning

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 22

参考文献 39

相关文章 15

编辑推荐

Metrics