The forecasting of the ionospheric $E_s$ layer critical frequency based on neural network computation techniques

The aim of this paper is to consider some problems of application of artificial neural networks for prediction of parameters of an ionosphere such as the sporadic Es layer critical frequency, $f_oE_s$, defined of electron's density the relevant heights. The character of interaction of radiowaves with an ionosphere depends on its value, therefore forecasting of $f_oE_s$ has essential applied significance. For prediction the artificial neural network is used. The choice of such approach is stipulated by that as in comparison with conventional methods of the forecasting the neural network is not bound to particular model of a predictablis phenomenon and the functional dependence is appeared during tutoring a network on the forecasting of particular parameter. As a result the neural network model of the forecasting of Es critical frequency is constructed, and the comparison with a conventional method of the forecasting operating the LS method is carried out. The results of comparison display better accuracy of the forecasting based on the use of artificial neural network.