Neural net clustering in the study of electrical grids failures in relation to geomagnetic storms

Abstract

Nowadays, the space weather issues are of a great importance, especially, for all satellites operators or ground-based electrical and electronic systems caused by geomagnetic storms. In spite of many studies on the changeable Sun and the fluctuations in the interplanetary space triggered by solar-driven disturbances, the question, which of the components of geomagnetic storms influence the strongest the electrical and electronic systems is still open. Here, we consider the data of electrical grids breakdowns having unidentified reasons, as well as failures connected to the aging of the infrastructure elements and breakdowns of electronic devices, which occurred during the periods of an increased geomagnetic activity. Mathematical and statistical methods, among them neural net clustering, demonstrate that there exist connections between the groups of the coefficients characterizing the state of Earths vicinity during magnetic storms and the number of failures in electrical network.