Sunspot group classification method based on deep learning

Abstract

The invention provides a sunspot group classification method based on deep learning. The classification method comprises the following steps: observing two images by using a sunspot group magnetic field and white light, and respectively processing the two images by using a deep learning network; performing certain weight processing on the prediction results of the two images according to the training precision of the deep learning model to obtain a comprehensive prediction result of the two images of the same sunspot group. Compared with an existing sunspot group classification method, the method utilizes two observation images of the sunspot group, predicts the same target through the deep learning network, comprehensively evaluates the predicted target according to the training precision of the two images, weights the prediction confidence, and further improves the classification precision of the sunspot group.

Description

technical field [0001] The invention relates to a method for classifying sunspot groups based on deep learning, and belongs to the technical fields of image processing and astronomy. Background technique [0002] The area where the magnetic field gathers on the surface of the sun’s photosphere is called sunspots, and when sunspots appear in groups in local areas, it is called sunspot groups. The appearance of sunspot groups is often associated with solar activities, and the appearance of some sunspot groups during solar activities, such as flares, will have a serious impact on the earth, disrupt the earth's ionosphere, generate magnetic storms, and disrupt wireless communications. Deep learning is a field of machine learning that has developed rapidly in recent years. Its concept comes from artificial neural networks, including multi-layer perceptrons with multiple hidden layers. The basic feature is to imitate the transmission between neurons in the brain, and the mode of ...

AI Technical Summary

Problems solved by technology

At present, in the method of classifying sunspot groups using deep learning, one type of observation image is used to train the model to realize the prediction of sunspot group, but different types of observation images are used to represent the characteristics of a certain sunspot group There are different accuracies on , therefore, there is still room for improvement in the accuracy of classifying sunspot groups

Images