Análise comparativa de redes neurais na previsão de intensidade de sinal de televisão digital na faixa de UHF (572 A 578 MHz) na cidade de Araguari

Abstract

This research presents a comparative analysis of Artificial Neural Networks based on predictive algorithms to estimate the electric field strength in digital terrestrial television signals in the range from 572 to 578 MHz, in the city of Araguari, Minas Gerais. The Levemberg-Marquardt, Bayesian Regularization and Scaled Conjugate Gradient algorithms had their results evaluated by the Regression Coefficient (R) and Mean Squared Error (MSE), the closer to 1 the R and the lower the MSE, better the results. The set of samples measured in the field was randomly divided into 80% for training, 10% for validation and 10% for testing. The three models provided high precision, and the Bayesian Regularization algorithm presented the best statistical indicators, both in its highest R regression value and in the lowest MSE, compared to the algorithms of Levemberg-Marquardt e Scaled Conjugate Gradient, respectively.