A Method for Forecasting Geomagnetic Storms Based on Deep Learning Neural Networks Using Time Series of Matrix Observations of the Uragan Muon Hodoscope

Abstract

A method for forecasting geomagnetic storms based on deep learning neural networks using digital time series processing for matrix observations of the URAGAN muon hodoscope and scalar Dst-indices has been developed. A scheme of computational operations and extrapolation formulas for matrix observations are proposed. The a variant of the neural network software module and its parameters are chosen. A decision-making rule is formed to forecast and assess the probabilities of correct and false forecasts of geomagnetic storms. An experimental study of estimates of the probabilistic characteristics and forecasting intervals of geomagnetic storms has confirmed the efficiency of the method. The obtained forecasting results are oriented towards solving a number of solar–terrestrial physics and national economic problems.