基于阶段式时序注意力网络的PM<sub>2.5</sub>鲁棒预测

陆瑶; 杨洁; 邵智娟; 朱聪聪

doi:10.13205/j.hjgc.202110013

基于阶段式时序注意力网络的PM_2.5鲁棒预测

doi: 10.13205/j.hjgc.202110013

陆瑶¹,
杨洁^1, ,,
邵智娟¹,
朱聪聪²

1. 苏州科技大学环境科学与工程学院, 江苏苏州 215009;
2. 上海大学计算机工程与科学学院, 上海 200444

基金项目:

国家自然科学基金面上项目（41571475）。

详细信息

作者简介:
陆瑶(1993-),女,硕士,主要研究方向为大气污染及其防治。Luyao0418@163.com

通讯作者:
杨洁(1972-),女,博士,教授,主要研究方向为环境管理与规划、大气污染及其防治。yjagnes@163.com

计量
- 文章访问数: 168
- HTML全文浏览量: 13
- PDF下载量: 9
- 被引次数: 0
出版历程
- 收稿日期: 2020-11-28
- 网络出版日期: 2022-01-26

PM_2.5 ROBUST PREDICTION BASED ON STAGED TEMPORAL ATTENTION NETWORK

1. College of Environmental Science and Engineering, Suzhou University of Science and Technology, Suzhou 215009, China;
2. College of Computer Engineering and Science, Shanghai University, Shanghai 200444, China

摘要

摘要: PM_2.5浓度的预测对于大气污染治理、改善环境质量等起到重要作用。受气象条件变化与大气污染物排放等多种因素的交叉影响，PM_2.5预测通常易受突变事件及噪声数据干扰。因此，基于对气象条件以及大气污染物与PM_2.5的相关性分析，提出阶段式时序注意力网络模型（staged temporal-attention network，STAN），该方法融合多段注意力学习模块与循环神经网络，建模气象因素与大气污染物对PM_2.5浓度的交叉影响。统计分析北京市、上海市、广州市预测结果的绝对误差值，可知：1）对比广泛使用的单一类模型支持向量机（support vector machine，SVM）、长短期时序记忆方法（long short-term memory，LSTM）和多层感知机（multilayer perceptron，MLP），STAN可达到10%以上的性能领先；对比最新的融合类模型U型网络（U-net），STAN领先了7%的优势。2）以北京市冬季预测结果为例进行统计分析，STAN的预测值与实测值之间的拟合系数可有95.2%的性能领先。此外，在鲁棒性分析中发现，STAN在含有10%噪声的数据上进行预测，误差上升幅度仅为9.3%。结果表明：注意力机制与时序学习模块相结合能够深度挖掘PM_2.5变化规律并抑制噪声数据，且STAN模型可以进行PM_2.5浓度的鲁棒预测。
- PM_2.5鲁棒预测 /
- 分段注意力学习 /
- 时序记忆网络 /
- 排放预警
Abstract: The forecast of PM_2.5 concentration plays an important role in air pollution control and improvement of environmental quality. Affected by multiple factors such as changes in meteorological conditions and air pollutants emissions, the PM_2.5 forecast was usually susceptible to sudden changes and noise data. Therefore, in-depth exploration of the PM_2.5 concentration change law and modeling robust prediction models have become key steps in this task. Based on the analysis of the correlation between meteorological conditions and atmospheric pollutants on PM_2.5, a staged temporal-attention network (STAN) was proposed. This method combined a multi-stage attention module and a recurrent neural network (RNN) to model the cross-influence of meteorological factors and atmospheric pollutants on PM_2.5 concentration. Statistical analysis of the absolute error values of the prediction results of Beijing, Shanghai, and Guangzhou showed the following results:1) compared with the widely used support vector machine (SVM), long short-term memory (LSTM), and multilayer perceptron (MLP), the performance of the proposed method increased more than 10%. 2) compared with the latest fusion model U-net, the proposed STAN still achieved a decrease in the error of 7%. Taking Beijing's winter forecast results as an example for statistical analysis, the fitting coefficient between the predicted value of STAN and the measured value could reach 95.2%. In the robustness analysis, it was found that the error increased by only 9.3% when the proposed method ran on data with 10% noise. The above results proved that the combination of the attention mechanism and the temporal learning module could deeply mine the change law of PM_2.5 and suppress noise data. It also showed that the STAN could achieve the robust prediction of PM_2.5 concentration.
- PM_2.5 robust prediction /
- staged attention learning /
- recurrent neural networks /
- emissions warning

HTML全文

参考文献(30)

[1]	张西雅,扈海波.基于多源数据的北京地区PM_2.5暴露风险评估[J].北京大学学报(自然科学版),2018,54(5):1103-1113.
[2]	CHAK K C, YAO X H. Air pollution in mega cities in China[J]. Atmospheric Environment,2008,42(1):1-12.
[3]	顾芳婷,胡敏,王渝,等.北京2009-2010年冬、春季PM_2.5污染特征[J].中国环境科学, 2016,36(9):2578-2584.
[4]	陈熙勐,张皓旻,顾万清,等.我国PM_2.5主要成分及对人体健康危害研究进展[J].中华保健医学杂志,2019,21(1):83-85.
[5]	于淼,郭立群,李秋爽,等.PM_2.5暴露下皮肤损伤评价方法研究进展[J].环境与职业医学,2019,36(9):869-873 ,878.
[6]	段肖肖,刘思秀.PM_2.5致病机制的研究进展[J].复旦学报(医学版),2020,47(4):605-614.
[7]	吴少伟, 邓芙蓉.大气PM_2.5与健康:从暴露、危害到干预的系统研究进展[J].中国药理学与毒理学杂志, 2016, 30(8):797-801.
[8]	王庚辰, 王普才.中国PM污染现状及其对人体健康的危害[J]. 科技导报, 2014,32(26):72-78.
[9]	KIM Y, MANLEY J, RADOIAS V. Medium-and long-term consequences of pollution on labor supply:evidence from Indonesia[J]. IZA Journal of Labor Economics, 2017,6(1):5.
[10]	BRAITHWAITE I, ZHANG S, KIRKBRIDE J B,et al.Air pollution (particulate matter) exposure and associations with depression, anxiety, bipolar, psychosis and suicide risk:a systematic review and meta-analysis[J]. Environmental Health Perspectives,2019, 127(12):126002.
[11]	KAN H D, CHEN R J, TONG S L. Ambient air pollution, climate change, and population health in China[J].Environment International,2012, 42:10-19.
[12]	陈波,鲁绍伟,李少宁.北京城市森林不同天气状况下PM_2.5浓度变化[J]. 生态学报, 2016,36(5):1391-1399.
[13]	王晓飞,王波,陆玉玉,等.基于Prophet-LSTM模型的PM_2.5浓度预测研究[J].软件导刊,2020,19(3):133-136.
[14]	王黎明,吴香华,赵天良,等.基于距离相关系数和支持向量机回归的PM_2.5浓度滚动统计预报方案[J].环境科学学报,2017, 37(4):1268-1276.
[15]	李建新,刘小生,刘静,等.基于MRMR-HK-SVM模型的PM_2.5浓度预测[J].中国环境科学,2019,39(6):2304-2310.
[16]	陈菊芬,李勇.基于多模态支持向量回归的PM_2.5浓度预测[J].环境工程,2019,37(1):122-126 ,34.
[17]	宋国君,国潇丹,杨啸,等.沈阳市PM_2.5浓度ARIMA-SVM组合预测研究[J].中国环境科学,2018,38(11):4031-4039.
[18]	赵文芳,林润生,唐伟,等.基于深度学习的PM_2.5短期预测模型[J].南京师大学报(自然科学版),2019,42(3):32-41.
[19]	黄婕,张丰,杜震洪,等.基于RNN-CNN集成深度学习模型的PM_2.5小时浓度预测[J].浙江大学学报(理学版),2019,46(3):370-379.
[20]	鲍艳磊,田冰,胡引翠,等.基于混合效应模型的京津冀地区PM_2.5数值模拟[J].环境污染与防治,2020,42(3):268-274.
[21]	黄俊,王超群,周宝琴,等.广州PM_2.5污染特征及与气象因子的关系分析[J].环境污染与防治,2020,42(2):176-181 ,186.
[22]	李定,张文生.面向群体行为识别的注意力池化机制[J].中国科学:信息科学,2021,51(3):399-412.
[23]	徐少伟,秦品乐,曾建朝,等.基于注意力机制的两阶段纵膈淋巴结自动分割算法[J].计算机应用,2021,41(2):556-562.
[24]	郭旭超,唐詹,刁磊,等.基于部首嵌入和注意力机制的病虫害命名实体识别[J].农业机械学报,2020,51(增刊2):335-343.
[25]	SANTOS C, MING T, BING X, et al. Attentive pooling networks[J]. 2016.
[26]	WILLIAMS R J,ZIPSER D.A learning algorithm for continually running fully recurrent neural networks[J]. Neural Computation, 1998,1(2).
[27]	HINTON G E, SRIVASTAVA N, KRIZHEVSKY A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J]. Computer Science, 2012,3(4):212-223.
[28]	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780.
[29]	RUCK D W,ROGERS S K,KABRISKY M,et al. The multilayer perceptron as an approximation to a Bayes optimal discriminant function[J]. IEEE transactions on neural networks,1990,1(4):296-298.
[30]	史晓涛,刘建丽,骆玉荣. 一种抗噪音的支持向量机学习方法[C]//全国第19届计算机技术与应用(CACIS)学术会议论文集(下册),2008:5.