Development of a Neural Network for Interpolation of Radio Wave Attenuation Coefficient Data

The article aims to study the possibility of replacing empirical models with corresponding neural network models. As an example, the empirical dependence of the attenuation coefficient of radio signals propagating at extremely low ambient temperatures is considered. To perform this task, an applied computer program has been developed that implements a simple neural network designed to recognize input signals and interpolate an empirical data sample. The program stores the results of training the neural network and performed calculations in a database, which allows you to effectively recognize input signals and select a suitable model. The results of a numerical experiment that allows us to evaluate the possibility of neural network modeling of the process of radio signal attenuation under the considered propagation conditions are presented.