一种基于小波变换和LSTM的大气污染物浓度预测模型

何哲祥; 李雷

doi:10.13205/j.hjgc.202103016

一种基于小波变换和LSTM的大气污染物浓度预测模型

doi: 10.13205/j.hjgc.202103016

何哲祥^,,
李雷

中南大学冶金与环境学院环境工程研究所, 长沙 410083

基金项目:

国家“十二五”科技支撑计划项目（2012BAC09B02）；中南大学中央高校基本科研业务费专项资金（2019zzts513）。

详细信息

作者简介:
何哲祥(1963-),男,博士,教授,主要研究方向为尾矿处置及污染控制、环境材料及矿山生态环境等。hezx@csu.edu.cn

通讯作者:
何哲祥(1963-),男,博士,教授,主要研究方向为尾矿处置及污染控制、环境材料及矿山生态环境等。hezx@csu.edu.cn

计量
- 文章访问数: 326
- HTML全文浏览量: 45
- PDF下载量: 16
- 被引次数: 0
出版历程
- 收稿日期: 2020-01-20
- 网络出版日期: 2021-07-19

AN AIR POLLUTANT CONCENTRATION PREDICTION MODEL BASED ON WAVELET TRANSFORM AND LSTM

HE Zhe-xiang^,,
LI Lei

Department of Environment Engineering, School of Metallurgy and Environment, Central South University, Changsha 410083, China

摘要

摘要: 针对现有大气污染物浓度预测模型存在预测精度不高、污染物种类单一等不足的问题，通过小波分解将高维大气污染物数据转换为低维数据，再对分解序列建立长短期记忆网络（LSTM）预测模型，最后通过小波重构将分解序列重构为污染物时间序列，建立了1种基于小波变换（WT）的LSTM大气污染物预测模型（WT-LSTM），用以预测目标区域内的次日平均ρ（PM_2.5）、ρ（PM₁₀）、ρ（SO₂）、ρ（NO₂）和ρ（O₃）。采用长沙市2015—2018年10处国控站点的数据进行验证，结果表明：相对于LSTM、多元线性回归（MLR）和基于WT的WT-MLR模型，WT-LSTM的均方根误差和绝对平均误差均下降了50%，其对PM_2.5、PM₁₀、SO₂、NO₂和O₃的污染等级预测准确率均在80%以上。
- 空气质量指数 /
- 机器学习 /
- 小波变换 /
- 大气污染物预测
Abstract: To solve the problem that current atmospheric pollutant prediction research has low accuracy of prediction and only pays attention to single pollutant type, a long short-term memory network atmospheric pollutant prediction model based on wavelet transform was proposed, to predict daily average PM_2.5, PM₁₀, SO₂, NO₂ and O₃ concentration of the next day. First, the high-dimensional data was converted into low-dimensional data by wavelet decomposition, and subsequently, the long short-term memory network prediction model was established for low-dimensional data. Finally, the decomposition sequence was reconstructed into the pollutant time series by wavelet reconstruction. Based on the data collected from 10 national control stations in Changsha from 2015 to 2018, the model was verified. The results showed that for the prediction of atmospheric pollutants of the next day, compared with the LSTM, MLR, WT-MLR, the root mean square error and absolute mean error of WT-LSTM model decreased by 50%, and the accuracy of the pollution level predictions of the five air pollutants were all above 80%.
- air quality index /
- machine learning /
- wavelet transform /
- air pollutant prediction

HTML全文

参考文献(23)

[1]	石佳超, 罗坤, 樊建人, 等. 基于CMAQ与前馈神经网络的区域大气污染物浓度快速响应模型[J]. 环境科学学报, 2018,38(11):4480-4489.
[2]	冯震, 李兴华. WRF-CMAQ模式在内蒙古本地化应用[J]. 内蒙古科技与经济, 2018(23):42-48,159.
[3]	赵文怡, 夏丽莎, 高广阔, 等. 基于加权KNN-BP神经网络的PM_2.5浓度预测模型研究[J]. 环境工程技术学报, 2019,9(1):14-18.
[4]	李晓理, 梅建想, 张山. 基于改进粒子群优化BP-Adaboost神经网络的PM_2.5浓度预测[J]. 大连理工大学学报, 2018,58(3):316-323.
[5]	孙宝磊, 孙暠, 张朝能, 等. 基于BP神经网络的大气污染物浓度预测[J]. 环境科学学报, 2017,37(5):1864-1871.
[6]	姚宁, 马青兰, 张晶, 等. 基于AGNES算法优化BP神经网络和GIS系统的大气污染物浓度预测[J]. 中国环境监测, 2015,31(3):113-117.
[7]	BAI Y, LI Y, WANG X X, et al. Air pollutants concentrations forecasting using back propagation[J]. Atmospheric Pollution Research, 2016,7(3):557-566.
[8]	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.
[9]	HU Z, LI W, QIAO J. Prediction of PM_2.5 based on Elman neural network with chaos theory[C]//201635th Chinese Control Conference (CCC). IEEE, 2016:3573-3578.
[10]	WANG P, LIU Y, QIN Z D, et al. A novel hybrid forecasting model for PM₁₀ and SO₂ daily concentrations[J]. Science of the Total Environment, 2015,505:1202-1212.
[11]	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):37,125 -129.
[12]	李颖若,汪君霞,韩婷婷, 等. 利用多元线性回归方法评估气象条件和控制措施对APEC期间北京空气质量的影响[J]. 环境科学, 2019,40(3):1024-1034.
[13]	PAK U, KIM C, RYU U, et al. A hybrid model based on convolutional neural networks and long short-term memory for ozone concentration prediction[J]. Air Quality, Atmosphere & Health, 2018, 11(8):883-895.
[14]	曲悦, 钱旭, 宋洪庆, 等. 基于机器学习的北京市PM_2.5浓度预测模型及模拟分析[J]. 工程科学学报, 2019,41(3):401-407.
[15]	WU L F, LI N, YANG Y J. Prediction of air quality indicators for the Beijing-Tianjin-Hebei region[J]. Journal of Cleaner Production, 2018,196:682-687.
[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]	ZHOU Q P, JIANG H Y, WANG J Z, et al. A hybrid model for PM_2.5 forecasting based on ensemble empirical mode decomposition and a general regression neural network[J]. Science of the Total Environment, 2014, 496:264-274.
[18]	QING T, FANG L, YONG L, et al. Air pollution forecasting using a deep learning model based on 1D convnets and bidirectional GRU[J].IEEE Access,2019,7:76690-76698.
[19]	陈国初, 王鹏, 徐余法, 等. 基于小波分解的风电场短期功率混合预测模型[J]. 上海电机学院学报, 2011,14(3):163-168.
[20]	黄婷婷,余磊. SDAE-LSTM模型在金融时间序列预测中的应用[J].计算机工程与应用,2019,55(1):142-148.
[21]	中国环境监测总站.全国城市空气质量实时发布平台[DB/OL]. http://106.37.208.233:20035/.
[22]	尹建光, 彭飞, 谢连科, 等. 基于小波分解与自适应多级残差修正的最小二乘支持向量回归预测模型的PM_2.5浓度预测[J]. 环境科学学报, 2018,38(5):2090-2098.
[23]	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.