MACHINE LEARNING METHOD FOR STRUCTURING COSMIC WEATHER FORECASTING SYSTEM, AND COSMIC WEATHER FORECASTING METHOD STRUCTURED BY THE METHOD

摘要

PROBLEM TO BE SOLVED: To structure a novel and useful cosmic weather forecast system by processing a large quantity of solar observation data.SOLUTION: A machine learning method is intended for structuring a cosmic weather forecast system comprising a first step of subjecting a whole sphere observation image of the sun observed at a time point tto two-dimensional Wavelet conversion to calculate the sum of all the wavelength components and thereby obtaining n-dimensional feature vectors and a following second step of machine-learning maps from a feature vector at the time point tintended for the object of cosmic weather forecasting observed from the time point tuntil a time point t+k. By implementing the first step and the second step at many time points, maps are machine-learned repeatedly.SELECTED DRAWING: Figure 3