A New Solar Flare Prediction Model based on Large-scale Image Management

摘要

Solar flare is one of the most powerful solar activities, which plays a very important role in daily life and space weather, so it is meaningful to predict solar flare accurately. In this paper, a new solar flare prediction model based on large-scale image management 'LIM-Sphere' has been proposed, which is built by large-margin, image indexing and hypersphere theory. Experimental results show that: 1) LM-Sphere has retained the distribution of original image data which reduced the accuracy loss of sampling in usual large-scale data modeling; 2) The application of large-scale image management method has solved the local minimum point's problem in K-nearest neighbor (KNN) algrithm which improved the accruacy of flare prediction. All in all, LVIsphere is a powerful model in solar flare prediction.