基于统计学习技术的太阳质子事件预报模型  

Solar Proton Prediction Model Based on Statistic Learning Technique

作　　者：李蓉[1] 黄鑫[2] 崔延美[3] 

机构地区：[1]北京物资学院,北京101149 [2]中国科学院国家天文台,北京100012 [3]中国科学院空间科学与应用研究中心,北京10090

出　　处：《科学技术与工程》2014年第28期5-8,16,共5页Science Technology and Engineering

基　　金：国家自然科学基金重点项目(11273031);北京市专业建设信息类特色专业建设项目(PXM2014_014214_000017)资助

摘　　要：结合太阳耀斑与日冕物质抛射参量作为预报因子建立太阳质子事件预报模型。描述太阳耀斑的三个特征参量包括耀斑峰值流量、持续时间和耀斑维度;太阳质子事件的三个特征参量分别为CME宽度、CME速度和测量位置角度。首先使用信息增益率评价各参量对质子事件发生的重要度,结果表明相比于耀斑峰值流量和持续时间,CME宽度和速度对质子事件发生具有更高的重要性。基于上述参量,应用线性Logistic回归方法建立质子事件预报模型。对模型进行检测并与只选用耀斑参量的预报模型的预报结果进行比较,结果显示采用耀斑结合CME参量的预报模型具有较高的预报准确率和较低的虚报率,尤其对于质子事件发生的报准率提高较多(67.5%上升到90%)。实验结果验证CME参量作为预报因子的有效性。Solar flare and coronal mass ejection parameters are chosen as predictors, a solar proton forecasting model is constructed. Solar flares are parameterized by the peak flux, the duration and the longitude, and CMEs are parameterized by the width, the speed and the measurement position angle （MPA）. The importance of each parameter is evaluated for the occurrence of SPEs by calculating the information gain ratio. The CME width and speed are shown more informative than the flare＇ s peak flux and duration. Based on these parameters, the Logistic regression method is employed to build the prediction model. The tests are done and the results are compared with the model base on only flare parameters. Results shows that the prediction model with connected parameters has a higher accuracy and lower false alarm rate than those of flare based one. Especially, the accuracy for SPEs occurrence raise much more obviously （from 67.5% to 90% ）. The test results validate that CME parameters are very effective predictors as well.

关 键 词：太阳质子事件 太阳耀斑 日冕物质抛射 信息增益率 LOGISTIC回归 

分 类 号：P182.9[天文地球—天文学]