基于主成分分析的 RBF 神经网络预测 SO2 浓度

大气与环境光学学报 ›› 2022, Vol. 17 ›› Issue (5): 550-557.

基于主成分分析的 RBF 神经网络预测 SO2 浓度

张琦锦1,2 , 郭映映1,2 , 李素文1,2∗ , 牟福生1,2∗

1 淮北师范大学物理与电子信息学院, 安徽淮北 235000; 2 污染物敏感材料与环境修复安徽省重点实验室, 安徽淮北 235000

收稿日期:2021-05-19 修回日期:2022-08-04 出版日期:2022-09-28 发布日期:2022-10-17
通讯作者: E-mail: swli@chnu.edu.cn;E-mail: moufusheng@163.com E-mail:qjzhang2020@outlook.com
作者简介:张琦锦 (1999 - ), 安徽阜阳人, 硕士研究生, 从事污染物监测与分析研究工作。 E-mail: qjzhang2020@outlook.com
基金资助:
Supported by National Natural Science Foundation of China (国家自然科学基金, 41875040), National Science Research Project of University in Anhui Province (安徽省高校自然科学研究项目, KJ2020A0029), Top-Notch Talents in University in Anhui Province (安徽省高校学科拔尖人才, gxbjZD2020067)

Prediction of SO2 concentration by RBF neural network based on principal component analysis

ZHANG Qijin 1,2 , GUO Yingying 1,2 , LI Suwen 1,2∗ , MOU Fusheng 1,2∗

1 School of Physics and Electronic Information, Huaibei Normal University, Huaibei 235000, China; 2 Anhui Province Key Laboratory of Pollutant Sensitive Materials and Environmental Remediation, Huaibei 235000, China

Received:2021-05-19 Revised:2022-08-04 Published:2022-09-28 Online:2022-10-17
Contact: Qi-Jin ZHANG E-mail:qjzhang2020@outlook.com

摘要/Abstract

摘要： 利用基于主成分分析 (PCA) 算法的径向基 (RBF) 神经网络对大气中 SO2 浓度进行滚动预测。以北京大兴地区 2019 年 9 月 1 日至 2020 年 10 月 31 日的气象数据和空气质量参数为基础, 结合逐步回归法筛选出与 SO2 线性相关的参数作为输入样本, 构建 PCA-RBF 预测模型。利用该模型预测北京大兴地区某天的 SO2 浓度, 将预测值保留并作为下一天预测模型的输入参数。以此将预测值不断地向前延伸并进行分析和预测, 从而实现 SO2 浓度的滚动预测。对比 RBF 网络和 PCA-RBF 网络两种模型的预测结果, 其中 PCA-RBF 模型期望值和预测值的误差及相关系数分别为 0.03 µg·m−3 和 0.9989。表明 PCA-RBF 网络模型能精准预测 SO2 浓度变化趋势, 为进一步解决大气污染问题提供技术支持。

关键词: 逐步回归分析, 主成分分析, 主成分分析-径向基神经网络, SO2

Abstract: The rolling prediction of atmospheric SO2 concentration using radial basis function (RBF) neural network based on principal component analysis (PCA) algorithm is presented. Based on the meteorological data and air quality parameters from September 1, 2019 to October 31, 2020 in Daxing district of Beijing, the PCA-RBF prediction model is constructed by combining the stepwise regression method to select the high correlation parameters between SO2 and meteorological factors as input samples. Then the PCA-RBF prediction model is used to predict the SO2 concentration in Daxing area on a certain day, and the prediction results are retained as the input parameters of the prediction model for the next day. In this way, the predicted value is continuously extended forward for the following analyzation and prediction, so as to realize the rolling prediction of SO2 concentration. Comparing the predicted results of RBF network and PCA-RBF network, the error and correlation coefficient of expected value and predicted value of PCA-RBF model are 0.03 µg·m−3 and 0.9989. It is shown that the PCA-RBF network model can accurately predict the variation trend of SO2 concentration, and provide a new technical support for further solving the air pollution problem.

Key words: stepwise regression analysis, principal component analysis, principal component analysis-radial basis function neural network, SO2

中图分类号:

X831

张琦锦, 郭映映, 李素文, ∗, 牟福生, ∗. 基于主成分分析的 RBF 神经网络预测 SO2 浓度[J]. 大气与环境光学学报, 2022, 17(5): 550-557.

ZHANG Qijin , , GUO Yingying , , LI Suwen , ∗ , MOU Fusheng , ∗. Prediction of SO2 concentration by RBF neural network based on principal component analysis[J]. Journal of Atmospheric and Environmental Optics, 2022, 17(5): 550-557.

参考文献 17

[1]	Jin J L, Ma J Z, Lin W L, et al. MAX-DOAS measurements and satellite validation of tropospheric NO2 and SO2 vertical
	column densities at a rural site of North China [J]. Atmospheric Environment, 2016, 133: 12-25.
[2]	Wu F C, Xie P H, Li A, et al. Investigations of temporal and spatial distribution of precursors SO2 and NO2 vertical columns
	in the North China Plain using mobile DOAS [J]. Atmospheric Chemistry and Physics, 2018, 18: 1535-1554.
[3]	Mou F S, Luo J, Li S W, et al. Measuring vertical profile and vertical column density of SO2 in troposphere by multi-axis
	differential optical absorption spectroscopy [J]. Acta Optica Sinica, 2019, 39(8): 0801001.
	牟福生, 雒静, 李素文, 等. 多轴差分吸收光谱技术测量对流层 SO2 垂直廓线及柱浓度 [J]. 光学学报, 2019, 39(8):
08	01001.
[4]	Qi H X, Guo Y Y, Mou F S, et al. Observation of HCHO before and after epidemic in Huaibei area based on MAX-DOAS [J].
	Acta Photonica Sinica, 2021, 50(01): 218-226.
	齐贺香, 郭映映, 牟福生, 等. 基于 MAX-DOAS 淮北地区疫情前后 HCHO 观测研究 [J]. 光子学报, 2021, 50(01): 218-226.
[5]	Jing R Y, Di Y H. Correlation analysis of air pollutant index and meteorological parmeters in summer [J]. Environment Engineering, 2018, 36(8): 83-87.
	景若愚, 狄育慧. 夏季空气污染物指数与气象参数相关性分析 [J]. 环境工程, 2018, 36(8): 83-87.
[6]	Park J, Chang S. A particulate matter concentration prediction model based on long short-term memory and an artificial neural
	network [J]. International Journal of Environmental Research and Public Health, 2021, 18(13): 6801.
[7]	Luna A S, Paredes M L L, et al. Prediction of ozone concentration in tropospheric levels using artificial neural networks and
	support vector machine at Rio de Janeiro, Brazil [J]. Atmospheric Environment, 2014, 98: 98-104.
[8]	Chen L, Ma G D. Study on wavelet analysis and neural network prediction of SO2 concentration in air [J]. Journal of Environmental Sciences, 2006, 26(9): 1553-1558.
	陈柳, 马广大. 大气中 SO2 浓度的小波分析及神经网络预测 [J]. 环境科学学报, 2006, 26(9): 1553-1558.
[9]	Huang H, Lan H Y, Huang Y B. A detection method of SO2 concentration based on DBN and ELM [J]. Journal of Atmospheric
	and Environmental Optics, 2020, 15(3): 207-216.
	黄鸿, 兰洪勇, 黄云彪. 基于深度信念网络和极限学习机的 SO2 浓度检测 [J]. 大气与环境光学学报, 2020, 15(3):
20	7-216.
[10]	Si Z J, Sun B S, Li X F. Prediction of air quality based on improved grey neural network model [J]. Chinese Journal of
	Environmental Engineering, 2013, 7(9): 3543-3547.
	司志娟, 孙宝盛, 李小芳. 基于改进型灰色神经网络组合模型的空气质量预测 [J]. 环境工程学报, 2013, 7(9): 3543-3547.
[11]	Dong J H, Wang G Y, Yao W. Similarity analysis model of water data based on theory of PCA [J]. Environmental Engineering,
20	16, 34(Sup1): 841-844.
	董建华, 王国胤, 姚文. 基于 PCA 的水质数据相似度分析模型 [J]. 环境工程, 2016, 34(S1): 841-844.
[12]	Zhang Y J. Research on PM2.5 Concentration Prediction Based on Principal Component Analysis and Neural Network [D].
	Qinhuangdao: Yanshan University, 2017.
	张亚吉. 基于主成分分析和神经网络的 PM2.5 浓度预测研究 [D]. 秦皇岛: 燕山大学, 2017.
[13]	Gao H S, Fu J M, Sun Y Q. Risk evaluation of fiber-optic relay protection channels based on PCA and RBF neural network
[J]	Journal of North China Electric Power University (Natural Science Edition), 2007,34(5): 90-94.
	高会生, 付建敏, 孙逸群. 基于 PCA-RBF 神经网络的光纤保护通道风险评估 [J]. 华北电力大学学报 (自然科学版),
20	07, 34(5): 90-94.
[14]	Zhang H. Research and Implementation of Air Pollution Monitoring and Forecasting System Based on LSTM Neural Network
[D]	Xi′
	an: Xidian University, 2019.
	张晗. 基于 LSTM 神经网络空气污染监测预报系统的研究与实现 [D]. 西安: 西安电子科技大学, 2019.
[15]	Miao B T, Chen F L. Applications of radius basis function neural networks in scattered data interpolation [J]. Journal of
	University of Science and Technology of China, 2001, 31(2): 135-142.
	缪报通, 陈发来. 径向基函数神经网络在散乱数据插值中的应用 [J]. 中国科学技术大学学报, 2001, 31(2): 135-142.
[16]	Sun B L, Sun H, Zhang C N, et al. Forecast of air pollutant concentrations by BP neural network [J]. Journal of Environmental
	Sciences, 2017, 37(5): 1864-1871.
	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.
[17]	Ouyang J, Wang A Z. The application of concentration forecasting of air pollutant based on BP neural network in Matlab [J].
	Environmental Science and Management, 2009, 34(11): 176-180.
	欧阳钧, 王爱枝. 基于 Matlab 的 BP 神经网络在大气污染物浓度预测中的应用 [J]. 环境科学与管理, 2009, 34(11):
17	6-180.