Analysis of square patch antennas for resonant frequency, return loss, gain and bandwidth using ANN

In this paper square patch antennas are modeled and analyzed using artificial neural network (ANN) having multi-layer perceptrons (MLP) network model. ANN used for analysis is first trained using data obtained by electromagnetic simulations of antennas using IE3D^?? software - based on full-wave analysis techniques. This model has four input parameters: Square patch antenna width W (patch antenna length L = W), substrate thickness h, relative permittivity of the substrate ??_r and probe feed position f_p - with respect to the centre along the width or length of the patch antenna. The ANN model has four output parameters: Resonant frequency f_r, return loss S₁₁, gain G and bandwidth BW. Levenberg-Marquardt backpropagation algorithm is used for training.