改进多元回归分析在空气质量监测的应用  

Application of Granger Causality and Multiple Regression Analysis in Air Quality Monitoring

作　　者：金江强 张怀相[1] 

机构地区：[1]杭州电子科技大学计算机学院,浙江杭州310018

出　　处：《杭州电子科技大学学报（自然科学版）》2016年第1期41-45,共5页Journal of Hangzhou Dianzi University：Natural Sciences

基　　金：国家科技支撑计划资助项目(2014BAF07B01)

摘　　要：为提高空气质量的测量精度,利用各种空气污染物之间的关联性,提出了一种基于空气污染物之间的因果关系来提高空气质量测量精度的算法.首先针对空气污染物的时间序列建立了自回归差分滑动平均模型;然后通过F统计量检验其格兰杰因果关系;接着利用逐步线性回归模型建立空气污染物之间的定量关系;最后运用实验数据分析并验证了算法的准确性和有效性.In order to improve the accuracy of measurement of air quality, this paper proposes an algorithm of improving air quality measurement precision by causality between air pollutants, based on the contact between the various air pollutants. First of all, autoregressive integrated moving average （AIMA） model with exogenous variables is established for time series of air pollutants. Secondly, Granger causality is tested for air pollutants by F-statistics. Then, stepwise linear regression mode is trained to establish a quantitative relationship in air pollutants which has causal relationship. Finally, the accuracy and effectiveness of the algorithm has been validated by the analysis of experimental data.

关 键 词：空气质量监测 无线传感网络 因果关系 多元回归 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]