基于支持向量机和k近邻的太阳质子事件预报模型

应用支持向量机和k近邻相结合的方法，建立了太阳质子事件预报模型。预报因子包括黑子面积、磁分型、McIntosh分型、太阳射电流量、活动区位置和软X射线流量。太阳质子事件模型包括两个子模型：质子有无模型和质子峰值流量模型。质子有无模型能对未来24小时是否发生质子事件给出预报，质子峰值流量模型对已发生的质子事件预报峰值流量等级。用2002年和2004年的数据进行了模拟预报，结果显示模型具有较高的报准率，同时显示出活动区位置和软x射线通量是比较敏感的预报因子。

Solar Proton Event Prediction Model Based on Support Vector Machine and k-Nearest Neighbors

The support vector machine(SVM)combined with the k-nearest neighbor(KNN)algorithm is applied to solar proton event prediction.The predictors of the model include sunspot area,magnetic class,Machtoish class,solar radio flux,active region location and soft X-ray flux are presented.The proton event prediction model is consisted of two sub-models.One is the proton occurrence model,the other is proton peak flux model.The first provides a solar event occurrence prediction,and the second provides a proton peak flux prediction for the predicted event occurrence.The verification for both models is made with testing data in 2002 and 2004.The testing results show that the soft X-ray flux and active region location are sensitive for the proton event occurrence and the SVM-KNN algorithm is an promising technique.