【摘要】 近年来我国多地区雾霾天气频发,针对PM_2.5浓度变化的非线性、时变性等特点,建立了基于支持向量机-小波神经网络（SVM-WNN）的组合预测模型。采用网格搜索算法对SVM的参数进行优化,利用优化后的模型进行初始预测,并结合WNN强大的非线性拟合能力的特点对其预测残差进行修正。以石家庄市每小时监测的PM_2.5浓度数据为样本建立模型并进行预测,结果表明,组合模型预测的平均相对误差为7. 2%。对比单一模型,组合模型的预测的效果更好,这也为短时PM_2.5浓度预测提供一个新的方法。更多还原

【Abstract】 In recent years,frequent haze weather has occurred in many regions in China is. In view of the nonlinear and timevarying characteristics of PM_2.5 concentration change,a combination prediction model based on support vector machine wavelet neural network（ SVM-WNN） is established. The grid search algorithm is used to optimize the parameters of the SVM,and the optimized model is used for initial prediction. The prediction residual is modified by combining the characteristics of WNN ’s strong nonlinear fitting ability. Based on the hourly monitoring data of PM_2.5 concentration in Shijiazhuang city,the model was established and predicted. The result showed that the average relative error of the combined prediction model is 7. 2%. Compared with the single model,the combined model is more effective in predicting PM_2.5 concentration,which provides a new method for short time PM_2.5 concentration prediction.更多还原