Increasing the metrological characteristics of anechoic chambers due to a posteriori analysis based on artificial neural networks

This article considers the possibility of improving the metrological characteristics of an anechoic chamber due to a posteriori processing of measurement results based on a generative adversarial model of an artificial neural network in order to reduce the influence on the distribution of the electromagnetic field in the measuring zone of waves reflected from the outer boundaries of the chamber and the equipment located in it. The training of the neural network was carried out on a data set obtained as part of a computational experiment and including the distribution of the electromagnetic field in the anechoic region for the model of an anechoic chamber and free space for given source layouts. The distributions of the real and imaginary parts of the electric component of the electromagnetic field were encoded with colour images. On the example of two-dimensional models of anechoic chambers, the practical feasibility of the proposed approach to a posteriori processing of measurement results is shown. Methods for estimating the accuracy of a posteriori processing of measurement results based on the metrics used to assess the quality of graphic images and calculating the errors in the amplitudes of the electric component of the electromagnetic field are given. The possibility of implementing the proposed method of a posteriori analysis in the framework of natural microwave measurements in anechoic chambers is assessed.