基于RBF神经网络的雾霾成因模型与预测

Establishment and Prediction of Haze Formation Model Based on RBF Neural Network

近些年,雾霾污染引起空气质量逐渐恶化,严重威胁人们的健康。以PM2.5为切入点分析雾霾的形成因素,提出一种基于径向基函数(RBF)神经网络建立的PM2.5雾霾成因模型,对污染源的主要因素进行分析。以武汉市空气质量数据为样本,开展非线性RBF雾霾成因模型的实证研究。研究结果表明,非线性RBF神经网络能够很好地拟合PM2.5与各成因变量之间的数量关系,获得了准确可靠的结果,为相关部门开展PM2.5的预测与分析提供了可靠的模型。

In recent years, haze pollution has not only caused the gradual deterioration of air quality, but also seriously threatened people’s health. In this paper, PM2.5 is taken as the key point of analyze the formation factors of haze. The formation model of PM2.5 haze is established based on RBF neural network and the main sources of pollution are found. Finally, the empirical study of the proposed nonlinear RBF model is carried out based on the data samples Wuhan city air quality. The results show that the nonlinear RBF neural network can match the quantitative relationship between PM2.5 and various genetic variables, and the accurate and reliable results were obtained. Hence, the proposed model could be provided for relevant department for the analysis and prediction of PM2.5.