Prediction of intensity of moderate and intense geomagnetic storms using artificial neural network during two complete solar cycles 23 and 24

This work aims to predict moderate, intense, and super geomagnetic storms during the two recent solar cycles 23 and 24 encompassing the period 1996–2018 using an artificial neural network (ANN). Optimization of the neural network includes a choice of activation function, training function, learning function, hidden layers, hidden neurons, learning rate, and momentum constant. The results obtained by the present study show the ability of the ANN model to produce an accurate estimate of the probability appearance of moderate and intense storms of about 88.9%.