基于CEEMDAN-SE-BiLSTM模型的西安市PM<sub>2.5</sub>浓度预测

谢琪; 夏飞; 袁博

doi:10.13205/j.hjgc.202408013

基于CEEMDAN-SE-BiLSTM模型的西安市PM_2.5浓度预测

doi: 10.13205/j.hjgc.202408013

谢琪¹,
夏飞^1, ,,
袁博²

1. 上海电力大学自动化工程学院, 上海 200090;
2. 南瑞集团有限公司(国网电力科学研究院有限公司), 南京 211106

基金项目:

上海市科委创新项目(19DZ1206800)

详细信息

作者简介:
谢琪(1998-),女,硕士研究生,主要研究方向为数据预测技术。xieqi9810@163.com

通讯作者:
夏飞(1978-),男,博士,副教授,主要研究方向为图像识别、数据挖掘。xiafeiblue@163.com

计量
- 文章访问数: 18
- HTML全文浏览量: 3
- PDF下载量: 1
- 被引次数: 0
出版历程
- 收稿日期: 2023-06-27
- 网络出版日期: 2024-12-02

PREDICTION OF PM_2.5 CONCENTRATION IN XI’AN BASED ON CEEMDAN-SE-BiLSTM MODEL

XIE Qi¹,
XIA Fei^{1
, ,},
YUAN Bo²

1. College of Automation Engineering, Shanghai University of Electric Power, Shanghai 200090, China;
2. NARI Group Corporation (State Grid Electric Power Research Institute), Nanjing 211106, China

摘要

摘要: 大气颗粒物浓度与环境污染息息相关,准确预测PM_2.5浓度对生态环境保护至关重要。选取西安市2020年1月1日—2021年12月31日的PM_2.5浓度数据和气象数据,针对PM_2.5序列非平稳非线性的特点,通过自适应噪声完备集合经验模态分解(CEEMDAN)将PM_2.5浓度序列分解为多个本征模态分量,减少数据噪声,以样本熵(SE)作为指标对分量进行k均值聚类(k-means聚类),去除冗余信息,然后将重构分量输入到双向长短期记忆神经网络模型(BiLSTM模型)中,辅以气象数据和独热编码处理后的时间数据增强输入特征,输出各分量的预测结果,叠加后得到最终的PM_2.5浓度预测结果。结果表明:与常见的XGBoost模型、长短期记忆神经网络(LSTM)模型、BiLSTM模型和其他组合模型相比,CEEMDAN-SE-BiLSTM模型在未来4个时刻(T+3、T+6、T+12、T+24)的预测性能更优。其均方根误差(RMSE)、平均绝对误差(MAE)、平均绝对百分比误差(MAPE)均有降低,T+3时刻的决定系数(R²)达到0.993,预测精度大幅提升。此外,在全国范围内选取5个城市(郑州、成都、北京、上海和广州)验证该模型的泛化性,结果显示,5个城市的预测误差均较小。CEEMDAN-SE-BiLSTM模型对PM_2.5浓度序列的短期预测具有较好的普适性、准确性。
- 双向长短期记忆神经网络(BiLSTM) /
- 自适应噪声完备集合经验模态分解(CEEMDAN) /
- PM_2.5浓度预测 /
- 样本熵(SE) /
- 时间序列
Abstract: Atmospheric particulate matter concentration is closely related to environmental pollution, and accurate prediction of PM_2.5 concentration is crucial for ecological environmental protection. Based on PM_2.5 concentration and meteorological data from January 1, 2020, to December 31, 2021, in Xi’an, the PM_2.5 concentration series was decomposed into multiple eigenmodal components by complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) for the non-stationary and non-linear characteristics, and the sample entropy (SE) was used as an indicator to perform k-means clustering to reduce data noise. Then the reconstructed components were inputted into a bi-directional long short-term memory model (BiLSTM model), supplemented with the enhanced information of meteorological data and temporal data after unique thermal coding processing to output the prediction results of each component, finally superimposed to obtain the final PM_2.5 concentration prediction results. The results showed that the CEEMDAN-SE-BiLSTM model had better prediction performance at four future moments (T+3, T+6, T+12, and T+24) compared with the XGBoost model, long short-term memory neural network (LSTM) model, BiLSTM model, and other combined models. The CEEMDAN-SE-BiLSTM model had better prediction performance, in terms of root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) were all decreased and at the moment of T+3, the determination coefficient (R²) was 0.993. The prediction accuracy was greatly improved. In addition, five cities (Zhengzhou, Chengdu, Beijing, Shanghai, and Guangzhou) were randomly selected nationwide to verify the model’s generalization, and the results showed that the prediction errors in the five cities were all small. The CEEMDAN-SE-BiLSTM model can be extended to other regions and cities and is capable of accurate short-term prediction.
- BiLSTM /
- CEEMDAN /
- PM_2.5 prediction /
- sample entropy /
- time series

HTML全文

参考文献(33)

[1]	WANG L F, SHI T, CHEN H Y. Air pollution and infant mortality: evidence from China[J]. Economics & Human Biology, 2023,49:101229.
[2]	GUO Q F, XI X, YANG S T, et al. Technology strategies to achieve carbon peak and carbon neutrality for China’s metal mines[J]. International Journal of Minerals, Metallurgy and Materials, 2022, 29(4):626-634.
[3]	秦思达, 惠秀娟, 夏广锋, 等. 基于Model-3/CMAQ模式的本溪市大气细颗粒物数值模拟[J].环境科学研究, 2018, 31(1):53-60.
[4]	秦思达, 王帆, 王堃, 等. 基于WRF-CMAQ模型的辽宁中部城市群PM_2.5化学组分特征[J].环境科学研究, 2021,34(6):1277-1286.
[5]	BHUVANESHWARI K S, UMA J, LOGESWARAN M N A, et al. Gaussian support vector machine algorithm based air pollution prediction[J]. CMC-Computers, Materials & Continua, 2022, 71(1):683-695.
[6]	谢永华, 张鸣敏, 杨乐, 等. 基于支持向量机回归的城市PM_2.5浓度预测[J]. 计算机工程与设计,2015, 36(11):3106-3111.
[7]	杨思琪, 赵丽华. 随机森林算法在城市空气质量预测中的应用[J]. 统计与决策, 2017(20):83-86.
[8]	任才溶,谢刚. 基于随机森林和气象参数的PM_2.5浓度等级预测[J]. 计算机工程与应用,2019,55(2):213-220.
[9]	彭斯俊, 沈加超, 朱雪. 基于ARIMA模型的PM_2.5预测[J].安全与环境工程,2014,21(6):125-128.
[10]	严宙宁, 牟敬锋, 赵星, 等. 基于ARIMA模型的深圳市大气PM_2.5浓度时间序列预测分析[J]. 现代预防医学,2018,45(2):220-223.
[11]	黄春桃, 范东平, 卢集富, 等. 基于深度学习模型的广州市大气PM_2.5和PM₁₀浓度预测[J]. 环境工程,2021,39(12):135-140.
[12]	ZHOU X X, XU J J, ZENG P, et al. Air pollutant concentration prediction based on GRU method[J]. Journal of Physics: Conference Series, 2019, 1168(3):032058.
[13]	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.
[14]	艾洪福, 石莹. 基于BP人工神经网络的雾霾天气预测研究[J]. 计算机仿真, 2015, 32(1):402-405 ,415.
[15]	陈建坤, 牟凤云, 张用川, 等. 基于多机器学习模型的逐小时PM_2.5浓度预测对比[J]. 南京林业大学学报(自然科学版), 2022, 46(5):152-160.
[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(1):997-1004.
[17]	KRISTIANI E, LIN H, LIN J R, et al. Short-term prediction of PM_2.5 using LSTM deep learning methods[J]. Sustainability, 2022, 14(2068):2068.
[18]	CHEN P, NIU A, JIANG W, et al. Air pollutant prediction: comparisons between LSTM, light GBM and random forests(Article)[J]. Journal of Environmental Protection and Ecology, 2019, 20(3):1059-1067.
[19]	LIU B C, GUO X L, LAI M Z, et al. Air pollutant concentration forecasting using long short-term memory based on wavelet transform and information gain: a case study of Beijing[J]. Computational Intelligence and Neuroscience, 2020(1):8834699.
[20]	YU L, WU C X, XIONG N N. An intelligent data analysis system combining ARIMA and LSTM for persistent organic pollutants concentration prediction[J]. Electronics, 2022, 11(4):652.
[21]	叶如珊, 王海波. 基于CNN-BiLSTM模型的PM_2.5浓度预测方法[J]. 数学的实践与认识, 2022, 52(7):181-188.
[22]	MOHAN A S, ABRAHAM L. An ensemble deep learning model for forecasting hourly PM_2.5 concentrations[J]. IETE Journal of Research, 2022:1-14.
[23]	MIRZADEH S M, NEJADKOORKI F, MIRHOSEINI S A, et al. Developing a wavelet-AI hybrid model for short- and long-term predictions of the pollutant concentration of particulate matter10[J]. International Journal of Environmental Science and Technology, 2022, 19(1):209-222.
[24]	HUANG Y, YU J H, DAI X H, et al. Air-quality prediction based on the EMD-IPSO-LSTM combination model[J]. Sustainability, 2022, 14(4889):4889.
[25]	WANG X W, LIU W J, WANG Y N, et al. A hybrid NO_x emission prediction model based on CEEMDAN and AM-LSTM[J]. Fuel, 2022, 310.
[26]	WANG J Q,TANG J L,GUO K, et al. Green bond index prediction based on CEEMDAN-LSTM[J].Frontiers in Energy Research,2022,9.
[27]	YANG Z Z,ZOU L,XIA J, et al. Inner dynamic detection and prediction of water quality based on CEEMDAN and GA-SVM models[J].Remote Sensing,2022,14(7):1714.
[28]	PENG R, KUN D, XU L, et al. Short-term load forecasting based on CEEMDAN and transformer[J]. Electric Power Systems Research,2023,214:108885.
[29]	LIU W F, JIANG Y, XU Y S. A super fast algorithm for estimating sample entropy[J]. Entropy, 2022, 24(4):524.
[30]	杨俊闯, 赵超. K-Means聚类算法研究综述[J]. 计算机工程与应用, 2019, 55(23):7-14.
[31]	SUN Y, LIU J W. AQI prediction based on CEEMDAN-ARMA-LSTM[J]. Sustainability, 2022,14(12182):12182.
[32]	DUAN H M, LUO X L. A novel multivariable grey prediction model and its application in forecasting coal consumption[J]. ISA Transactions, 2022, 120:110-127.
[33]	郑霞, 胡东滨,李权. 基于小波分解和SVM的大气污染物浓度预测模型研究[J]. 环境科学学报, 2020, 40(8):2962-2969.