ИСТИНА |
Войти в систему Регистрация |
|
Интеллектуальная Система Тематического Исследования НАукометрических данных |
||
Using artificial neural network (ANN) of Kohonen layer type, images of causal relationship of substorm activity with the Solar wind and interplanetary magnetic field parameters corresponding to the solar fluxes of magnetic cloud (MC) type within its influence on Earth’s magnetosphere were classified. The study was performed using minute data corresponding to the observation intervals of 33 MC’s, recorded in 1998–2012 [Barkhatov et al., 2014]. Solar wind parameters, interplanetary magnetic field components (IMF), values of SYM/H and AL indices for magnetic activity were analyzed for each MC interval. Based on available data, an information database was created, which contains 34 parameters. The analysis of classification results allows identifying the selected classes of substorms with a specific combination of disturbances of the solar wind parameters and IMF in Solar wind magnetic clouds. The implementation of machine vision algorithms was to develop a data presentation form for training and testing the ANN. For these purposes, combinations of parameters involved in the classification were presented in the form of three-, four-, etc. polygons. This way of presenting data arrays made it possible to monitor the ANN work and calculate the classification success. The classification experiments were carried out with separate use of parameters combinations that correspond only to the events causes (parameters related to the MC) and only the events consequences (parameters related to the geomagnetic response of the magnetosphere). The resulting classes of causes and effects classes were compared by a special algorithm. In the case of coincidence of causal parameters combinations class with the substorm investigation class, the class was declared established. As a result, each notable substormsclass was identified with a specific type of Solar wind and IMF parameters disturbance within MC body. Class 1 - manifestations in AL index dynamics in the form of solitary weak substormswith slowly changing values of Bz component within MC body. Class 2 - moderate manifestations of substorm activity in AL index dynamics in the form of solitary substorms or substorms series caused by sharp changes in Bzwithin MC body. Class 3 - extreme manifestations of the substorm activity in the form of a substormsseries with extreme AL index values, accompanied by a significant growth rate of the integral value NV2within MC body. It is shown that the use of parameters combinations as ANN input allows determining levels for expected AL index intensity with accuracy up to ~70%. The success of identifying specific cause-effect classes containing parameters of substorm activity causes and its dynamics indicates a close non-linear relationship between AL index dynamics and MC parameters. This work was supported by Ministry of Education and Science of Russian Federation project №5.5898.2017/8.9 (Barkhatov N.A.,Revunov S.E.). References Barkhatov N. A., Revunova E. A., Vinogradov A. B. Effect of Orientation of the Solar Wind Magnetic Clouds on the Seasonal Variation of Geomagnetic Activity // Cosmic Research. 2014. V. 52. N. 4. P. 269–277. © Pleiades Publishing, Ltd., 2014. DOI: 10.1134/S0010952514040017