时间序列模型预测大气臭氧浓度  

Atmospheric Ozone Concentration Prediction by Using Time Series Model

作　　者：王一龙 董韶妮 孙丽萍 王上 WANG Yilong;DONG Shaoni;SUN Liping;WANG Shang(Zhifu Environmental Monitoring Center of Yantai,Yantai 264000,Shandong,China;Yantai Environmental Monitoring Center,Yantai 264003,Shandong,China;School of Civil Engineering,Yantai University,Yantai 264005,Shandong,China)

机构地区：[1]烟台市芝罘环境监控中心,山东烟台264000 [2]烟台市环境监控中心,山东烟台264003 [3]烟台大学土木工程学院,山东烟台264005

出　　处：《济南大学学报（自然科学版）》2023年第2期178-183,共6页Journal of University of Jinan(Science and Technology)

基　　金：国家自然科学基金项目(52008362);山东省自然科学基金项目(ZR2019PD003)。

摘　　要：为了给大气污染防治预警预报提供参考,利用时间序列模型对大气臭氧浓度进行预测;以2020年1月1日至2020年12月31日期间366个烟台市区大气臭氧日均质量浓度作为研究数据,建立自回归移动平均模型,引入广义自回归条件异方差模型消除时间序列自回归条件异方差效应,最终构建自回归移动平均-广义自回归条件异方差时间序列模型,并对2021年1月烟台市区的大气臭氧日均浓度进行预测。结果表明,所构建的时间序列模型对大气臭氧浓度的短期预测值与实测值基本一致,但随着预测期数的增加,预测值与实测值的相对误差逐渐增大。To provide reference for early warning and forecasting of air pollution prevention and control, atmospheric ozone concentrations were predicted by using time series model. Taking daily average ozone concentrations of 366 Yantai urban areas from January 1, 2020 to December 31, 2020 as research data, autoregressive moving average model was established. Generalized autoregressive conditional heteroskedasticity model was introduced to eliminate autoregressive conditional heteroscedasticity effect of time series. Autoregressive moving average-generalized autoregressive conditional heteroskedasticity model was finally built, which was used to predict daily average concentrations of atmospheric ozone in Yantai urban areas in January, 2021. The results show that the short-term predicted values of atmospheric ozone concentrations by using the built time series model are basically consistent with the measured values, but relative errors between the predicted values and the measured values gradually increase with the increase of prediction period number.

关 键 词：臭氧浓度预测 时间序列模型 自回归移动平均模型 广义自回归条件异方差模型 

分 类 号：X803.3[环境科学与工程—环境工程]