Evolving Recurrent Neural Models of Geomagnetic Storms

Abstract

Genetic algorithms for training recurrent neural networks (RNNs) have not yet been considered for modeling the dynamics of magnetospheric plasma. We provide a discussion of the previous state of the art in modeling  D _st . Then, a recurrent neural network trained by a genetic algorithm is proposed for geomagnetic storm forecasting. The exogenous inputs to the RNN consist of three parameters, $\mbox{\boldmath$ b_z$}$, $\mbox{\boldmath$ n$}$, and $\mbox{\boldmath$ v$}$, which represent the southward and azimuthal components of the interplanetary magnetic field ( IMF), the density of electromagnetic particles, and the velocity of the particles respectively. The proposed model is compared to a model used in operational forecasts on a series of geomagnetic storms that so far have been difficult to forecast. It is shown that the proposed evolutionary method of training the RNN outperforms the operational model which was trained by gradient descent.