Design, Optimization and Prediction of the Performances for a Multi-band Patch Antenna Using Artificial Neural Network (ANN)

In this contribution, a rectangular U-slotted Patch antenna was designed and optimized using artificial neural network (ANN) based on the multilayer perceptron (MLP) model. First, the training data set was generated using MATLAB and HFSS software then an Artificial Neural Network Toolbox of MATLAB has been used for the training and the validation of the ANN. To check its accuracy, a comparison between the ANN resonant frequencies and the HFSS simulated frequencies was conducted. From then, it was concluded that the MLP is able to predict all the resonant frequencies with outstanding precision as the results obtained through these algorithms show very good agreement with the available experimental results. Moreover, the optimized antenna operates in three different frequencies and shows a very low reflection coefficient value for each resonant frequency (less than −30 dB), with good performances in terms of: gain, radiation pattern, and VSWR.