季节指数改进的PM2.5质量浓度组合预测模型研究

doi:10.12052/gdutxb.210055

摘要/Abstract

摘要： 随着我国的经济和城市化迅速发展，PM_2.5主导的区域空气污染已成为紧迫、突出的环境问题。据相关研究表明，PM_2.5在不同季节质量浓度差异较大。根据广州市2015~2019年的PM_2.5月均质量浓度数据，结合大气污染物及气象因素，引入季节指数，建立预测PM_2.5质量浓度的改进多元线性回归和多层感知器组合预测模型，探析广州市大气污染物中PM_2.5质量浓度的变化规律。结果表明，用季节指数改进的组合预测模型对PM_2.5质量浓度进行预测分析，拟合结果良好。使用不同评价指标将组合模型与传统的多层感知器预测模型和多元线性回归模型进行对比，该组合模型的均方根误差(Root Mean Square Error，RMSE)、平均绝对百分比误差(Mean Absolute Percentage Error，MAPE)、平均绝对误差(Mean Absolute Error，MAE)分别比多层感知器模型减少了23.1%、31%、24.2%；比多元线性回归模型减少了35.3%、41.3%、41%。该模型精度均优于传统的多元线性回归模型和多层感知器模型，能更好地预测环境PM_2.5质量浓度，为优化环境提供参考。

关键词: PM_2.5质量浓度预测, 多元线性回归, 多层感知器, 神经网络

Abstract: In recent years, China's economy and urbanization have developed rapidly, and the development of cities often comes at the expense of the environment. Regional air pollution dominated by PM_2.5 has become the most pressing and prominent environmental problem in China. According to relevant studies, PM_2.5 concentrations vary greatly in different seasons. Based on the PM_2.5 monthly mean concentration data of Guangzhou from 2015 to 2019, combined with atmospheric pollutants and meteorological factors, with the seasonal index introduced, an improved multiple linear regression and multi-layer perceptual combination prediction model for PM_2.5 concentration is established, to analyze the variation law and future development trend of PM_2.5 concentration in Guangzhou. The results show that the combined prediction model with the improved seasonal index is used to predict and analyze PM_2.5, and the fitting results are good. Compared with the multi-layer perceptron prediction model by using different evaluation indexes, the RMSE(Root Mean Square Error), MAPE(Mean Absolute Percentage Error) and MAE(Mean Absolute Error) of the combined model are reduced by 23.1%, 31% and 24.2% . Compared with the multiple linear regression model, the reduction is 35.3%, 41.3% and 41% . The accuracy of the model is better than the traditional multiple linear regression model and multi-layer perceptron model, which can better predict environmental PM_2.5 concentration and provide reference for optimizing the environment.

Key words: PM_2.5 concentration forecasting, multiple linear regression, multilayer perceptron, neural networks

中图分类号:

X831

曾江毅, 李志生, 欧耀春, 金宇凯. 季节指数改进的PM_2.5质量浓度组合预测模型研究[J]. 广东工业大学学报, 2022, 39(03): 89-94.

Zeng Jiang-yi, Li Zhi-sheng, Ou Yao-chun, Jin Yu-kai. PM_2.5 Concentration Improving Prediction Modeling of Seasonal Index[J]. Journal of Guangdong University of Technology, 2022, 39(03): 89-94.

参考文献

[1] FAN H, ZHAO C F, YANG Y K. A comprehensive analysis of the spatio-temporal variation of urban air pollution in China during 2014–2018 [J]. Atmospheric Environment, 2020, 220(117066): 1-12.
[2] 王玮, 汤大钢, 刘红杰, 等. 中国PM_2.5污染状况和污染特征的研究[J]. 环境科学研究, 2000(1): 1-5.
WANG W, TANG D G, LIU H J, et al. Research on corrent pollution sataus and pollution characteristics of PM_2.5 in China [J]. Research of Environmental Sciences, 2000(1): 1-5.
[3] 甘阳阳, 李志生. 空气幕对厨房内PM_2.5控制效果的模拟与分析[J]. 广东工业大学学报, 2020, 37(3): 82-87.
GAN Y Y, LI Z S. A numerical effect on PM_2.5 in the kitchen [J]. Journal of Guangdong University of Technology, 2020, 37(3): 82-87.
[4] 张洪. 京津冀地区空气质量时空分布特征及PM_2.5浓度预测[D]. 哈尔滨: 哈尔滨师范大学, 2020.
[5] XU X D, TONG T, ZHANG W, et al. Fine-grained prediction of PM_2.5 concentration based on multisource data and deep learning [J]. Atmospheric Pollution Research, 2020, 11(10): 1728-1737.
[6] MASOOD A, AHMAD K. A model for particulate matter PM_2.5 prediction for Delhi based on machine learning approaches [J]. Procedia Computer Science, 2020, 167(1): 2101-2110.
[7] 张怡文, 郭傲东, 吴海龙, 等. 基于PCA-BP神经网络的PM_2.5季节性预测方法研究[J]. 南京林业大学学报(自然科学版), 2020, 44(5): 231-238.
ZHANG Y W, GUO A D, WU H L, et al. Seasonal prediction of PM_2.5 based on the PCA-BP neural network [J]. Journal of Nanjing Forestry University (Natural Sciences Edition), 2020, 44(5): 231-238.
[8] HE H D, LI M, WANG W L, et al. Prediction of PM_2.5 concentration based on the similarity in air quality monitoring network [J]. Building and Environment, 2018, 137(1): 11-17.
[9] ASHA B C. Estimating PM_2.5 concentration from satellite derived aerosol optical depth and meteorological variables using a combination model [J]. Atmospheric Pollution Research, 2019, 10(3): 847-857.
[10] LIU Z R, HU B, WANG L L, et al. Seasonal and diurnal variation in particulate matter (PM ₁₀ and PM _2.5 ) at an urban site of Beijing: analyses from a 9-year study [J]. Environmental Science and Pollution Research, 2015, 22(1): 627-642.
[11] 李梓铭, 孙兆彬, 邵勰, 等. 北京城区PM_2.5不同时间尺度周期性研究[J]. 中国环境科学, 2017, 37(2): 407-415.
LI Z M, SUN Z B, SHAO X, et al. Using Morlet wavelet analysis to analyze multiple time scale periodically in PM_2.5 in Beijing [J]. China Environmental Science, 2017, 37(2): 407-415.
[12] 广东省统计局. 广东省第七次全国人口普查公报[EB/OL]. (2021-05-15)[2021-05-30]. http://stats.gd.gov.cn/attachment/0/421/421374/3284452.pdf.
[13] 何丹, 徐宝平, 吴丽芳, 等. 江西省典型城市气象因素对PM_2.5和PM₁₀污染特征影响分析——以萍乡市为例[J]. 能源研究与管理, 2020(4): 42-46.
HE D, XU B P, WU L F, et al. Analysis of the impact of meteorological factors on the pollution characteristics of PM_2.5 and PM₁₀ in the typical cities of Jiangxi Province—taking Pingxiang city as an example [J]. Energy Research and Mangement, 2020(4): 42-46.
[14] 焦晨, 黄艳然, 赵钦风, 等. 时间序列分解模型在山东省糖尿病门诊量预测中的应用[J]. 中国农村卫生事业管理, 2021, 41(2): 93-97.
JIAO C, HUANG Y R, ZHAO Q F, et al. Application of time series decomposition model to the predictionof the number of diabetics in Shandong Province [J]. Chinese Rural Health Service Administration, 2021, 41(2): 93-97.
[15] 郑薇, 王灿强, 李维德. 基于季节指数调整与HGWO-SVR算法的农产品价格预测模型[J]. 统计与决策, 2018, 34(19): 33-37.
ZHENG W, WANG C Q, LI W D. Forecasting model of agricultural products prices based on seasonal index adjustment and HGWO-SVR algorithm[J]. Statistics Decision, 2018, 34(19): 33-37.
[16] 郭秀英, 尹兴国, 张艳云. 季节性波动时间序列预测的分解季节指数法[J]. 数理统计与管理, 2000(6): 60-64.
GUO X Y, YIN X G, ZHANG Y Y. The analysis method of seasonal index number of time sequence of seasonal fluctuation [J]. Journal of Applied Statistics and Management, 2000(6): 60-64.
[17] XU W, REN G, LIU Y, et al. Forecasting energy consumption using a new GM–ARMA model based on HP filter: the case of Guangdong Province of China [J]. Economic Modelling, 2015, 45(1): 127-135.
[18] 陈通箭. 基于改进小波神经网络的高速公路季节性车流量预测模型[J]. 智能城市, 2020, 6(3): 22-24.
CHEN T J. Seasonal traffic flow prediction model of expressway based on improved wavelet neural network[J]. Intelligent City, 2020, 6(3): 22-24.
[19] SUN S, RAPPAPORT T S, RANGAN S, et al. Propagation path loss models for 5G urban micro-and macro-cellular scenarios[C]// 2016 IEEE 83rd Vehicular Technology Conference (VTC Spring). Nanjing: IEEE, 2016: 1-6.
[20] 谢劭峰, 赵云, 李国弘, 等. 基于WPA-WOA-BP神经网络的PM_2.5浓度预测[J]. 大地测量与地球动力学, 2021, 41(1): 12-16.
XIE S F, ZHAO Y, LI G H, et al. Prediction of PM_2.5 concentration based on WPA-WOA-BP neural network [J]. Journal of Geodesy and Geodynamics, 2021, 41(1): 12-16.
[21] 王徐凯, 余华银, 杨桂元, 等. 基于多层组合模型的武汉市月平均PM_2.5浓度预测[J]. 牡丹江师范学院学报(自然科学版), 2021(1): 7-11.
WANG X K, YU H Y, YANG G Y, et al. Prediction of nonthly average PM_2.5 concentration based on multilayer combination model: a case study of Wuhan [J]. Journal of Mudanjiang Normal University(Natural Sciences Edition), 2021(1): 7-11.
[22] 郭骐嘉, 姚宜斌, 周永江. 融合GNSS气象参数的PM_2.5随机森林预测模型[J]. 测绘科学, 2021, 46(4): 37-42.
GUO Q J, YAO Y B, ZHOU Y J. PM_2.5 random forest prediction model incorporating GNSS meteorological parameters[J]. Science of Surveying and Mapping, 2021, 46(4): 37-42.
[23] 王赛兰. 基于多元方法和时间序列的PM_2.5浓度分析与预测[D]. 长沙: 湖南师范大学, 2019.
[24] 魏朋志, 谢劭峰, 周志浩, 等. 节假日效应下的PM_2.5浓度预测[J/OL]. 桂林理工大学学报, 2021: 1-7(2021-03-03)[2021-03-22]. http://kns.cnki.net/kcms/detail/45.1375.N.20210303.1307.004.html.
WEI P Z, XIE S F, ZHOU Z H, et al. PM2.5 concentration prediction under holiday effect [J/OL]. Journal of Guilin University of Technology, 2021: 1-7 (2021-03-03)[2021-03-22]. http://kns.cnki.net/kcms/detail/45.1375.N.20210303.1307.004.html.
[25] 陈兵红, 靳全锋, 柴红玲, 等. 浙江省大气PM_2.5时空分布及相关因子分析[J]. 环境科学学报, 2021: 41(3): 817-829.
CHEN B H, JIN Q F, CHAI H L, et al. Spatiotemporal distribution and correlation factors of PM_{2. 5}concentrations in Zhejiang Province[J]. Acta Scientiae Circumstantiae, 2021: 41(3): 817-829.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed