Number of Dipoles for Electromagnetic Source Modeling Using Genetic Algorithm

The Far-Field distribution of an unsymmetrical printed circuit board can be approximated by an equivalent set of dipoles. Genetic Algorithm is employed to optimize the parameters of the dipoles using Near-Field data. The minimum number of dipoles required to represent the radiation source depends upon the expected complexity of the Far-Field radiation pattern. In general, with the increase in frequency, the Far-Field pattern becomes more complex and Genetic Algorithm requires more number of dipoles to represent the source. However, it is not necessary that at a low frequency, Genetic Algorithm would always be able to find the parameters of small number of dipoles required for the Far-Field approximation of the source. This relationship between the frequency and the number of dipoles required for Near-Field to Far-Field conversion using Genetic Algorithm is considered in this paper. The Far-Field plot of the printed circuit board predicted by Genetic Algorithm is investigated at different frequencies.