支持向量机应用于大气污染物时间序列预测

Support Vector Machine Applied to Air Pollutant Time Series Forecasting

阐述了支持向量机应用于大气污染物时间序列预测的具体方法,建立了大气污染物时间序列的支持向量机预测模型。该方法将支持向量机应用于大气污染物浓度预测:首先通过选择合适的信息量准则来确定模型阶数;而后通过实验的方法选择参数从而形成支持向量机的训练样本集,在此基础上建立了基于支持向量机的时间序列大气污染预测模型。实例表明,无论是在仿真过程还是在预测过程,支持向量机都具有很高的预测精度。因此,采用支持向量机方法对大气污染物时间序列进行预测分析是可行的。

The specific method of support vector machine （SVM） applied to air pollutant time series forecasting is expounded and a SVM prediction model of the air pollutant time series is constructed. SVM used in the concentration prediction of air pollutant by the method, first selecting suitable information criteria to determine the order of the model, then choosing parameters by experiment to form the training sample set of SVM, and then the time series air pollution prediction model has been created based on SVM. The example shows that the prediction accuracy of SVM is high whatever simulation or forecasting process. So using SVM to predict and analyze the air pollutant time series is feasible.