Resonance frequency of Vertical SIW based on the artificial neural network

A multilayers vertical substrate integrated waveguide (VSIW) by etching coupling aperture on each copper layer is presented in this paper. It is a promising bandpass transmission structure when apertures are in good vertical alignment. The apertures randomly shift along the horizontal plane leads to a stopband phenomenon with unwanted TM
110
mode resonance. Traditional three-dimension (3-D) simulation is not suitable for infinite geometries simulation. In our proposed method, geometries of VSIW are taken as input, and resonance frequencies are taken as the output of the artificial neuron network (ANN). We use the powerful mapping ability of ANN to replace the 3-D simulation so that finding resonance frequencies of corresponding geometries is effective. 7901 samples are obtained and randomly selected 70% and 30% are used to establish two ANNs, respectively. The resonance frequencies of the rest randomly selected 15% and 50% geometries predicted by two ANNs are almost the same as the simulation results. This shows the effectiveness of the proposed ANN method. This paper provides a new idea for mass time-consuming simulations.