回归分析与基于MIV的RBF神经网络在PM2.5的相关因素分析中的应用

The Application of Regression Analysis and RBF Neural Network Based on MIV in the Related Factors Analysis of PM2.5

PM2.5作为大气首要污染物,严重影响着人们的身体健康.为了研究影响PM2.5的相关指标,以武汉市的空气数据为研究对象,通过多元线性回归、偏最小二乘回归、基于MIV的RBF神经网络回归等方法对AQI中6个基本监测指标的PM2.5(含量)与其它5项分指标及其对应污染物(含量)之间的相关性进行分析;通过比较,基于MIV的RBF神经网络回归模型拟合度达到0.9302,效果最好,而且也优于BP人工神经网络回归算法,因此得出了精确可靠的影响PM2.5的指标权重大小,为减排PM2.5提供了可靠的理论依据.

PM2.5, as primary atmospheric pollutants, seriously affects people＇s health. In order to study the related indicators of PM2.5, this paper researches air data set of Wuhan city, through multiple linear regression, partial least squares regression and RBF neural network regression based on MIV, in six basic monitoring index of AQI, Correlation analysis is researched among PM2.5 （content） and the corresponding pollutants （content） of other five indicators; By comparison, the model fitting degree of RBF neural network regression based on MIV is 0.9302,and the result is not only best, but also superior to the BP artificial neural network regression algorithm. The accurate and reliable weight size is obtained about influencing indicators of PM2.5, this paper provides the reliable theory basis for the emission reduction of PM2.5.