Evaluation of radio propagation parameters for field strength prediction using neural network

This work utilizes artificial neural network to determine the electrical characteristics of the ground. This neural simulator is trained through the field strength of the surface wave, using the smooth spherical earth propagation method equations. To obtain the electric parameters of the ground, a campaign of measurements of the received electric field strength was carried out in around of Brasilia city. A radio station transmitter was used operating in medium wave (MW). The results of the electric parameters of the ground were found as: 4.23 mS/m for conductivity and 15 for relative permittivity. These results allows to conclude that the electrical characteristics of the ground described in the ITU-R P. 832-2 Recommendation, "World Atlas of Ground Conductivities" cannot translate the reality for the area studied in Brasilia.