基于多元线性回归算法的雾霾预测模型的研究

Study on Smog Prediction Model Based on Multiple Linear Regression Algorithm

对乌鲁木齐市环境监测站2013—2015年冬季逐日AQI、PM_(2.5)、PM_(10)、SO_2、NO_2、CO、O_3数据进行相关分析,并利用MATLAB编程工具进行多元回归统计分析,建立了多元回归统计预测模型。对2015年1—3月乌鲁木齐雾霾天气进行预测试验,发现预测值与实际值有较好的拟合效果和预报效果。实验证明,在大气层结稳定的冬季将当天的大气污染物浓度作为因子,用多元线性回归算法建立预测模型对次日雾霾天气进行预测是一种有效的雾霾统计预报手段,本文试图用MATLAB编程工具建立动态多元回归预测模型,编写了自动预测系统软件,测试取得了较好的预测效果。

In this paper, the daily data of AQI, PM2.5, PM10, SO2, NO2, CO and O3 observed in Urumqi Environmental Monitoring Station in winter from 2013 to 2015 was analyzed, and the multivariate regression statistical prediction model was established by using MATLAB programming tools for multivariate regression statistical analysis. The smog prediction experiment of Urumqi city from January to March in 2015 showed that the predicted values fitted well with the actual values. The study showed that, in winter with stable atmosphere, the Smog prediction model using the multiple linear regression algorithm was an effective means of forecasting Smog weather when the exact atmospheric pollutant concentration was taken as a main multiple linear regression factor. This paper attempted to establish a dynamic multiple regression Smog prediction model by using MATLAB programming tool, and compiled the automatic Smog prediction software. The Smog prediction experiment has achieved the good prediction results.