Investigation of the resonance frequence and half– powerbeamwidth of microstrip patch antennas by usingartificial neural network method

Microstrip antennas have many advantages such as easy fabrication, low cost and small size design. So, they are preferred in many applications like wireless communications, satellite and radar systems. Microstrip antennas can be designed in many different shapes; square, rectangular and circular. In this study, it is aimed to investigate the resonance frequency and the half– power beamwidth of microstrip patch antennas using artificial neural network (ANN) method. In the study, H-shaped microstrip patch antenna design, simulation, ANN modeling and production were performed. ANN, which is a deterministic analysis method, was preferred because it can be applied to many different problems and gives successful results. It prevents loss of time during designing of microstrip antennas, processing can be done at lower cost, production can be done in a shorter time and mathematical operations is not required. In our study, an antenna design with 390 different parameters had been performed and analyzed for use in model development. ANN model was performed by using Levenberg Marquart and Bayesian Regularization training algorithms, single and multi-layer model types and different neuron numbers. The results were evaluated, the model with the most suitable parameters was preferred for antenna design and it was produced. As a result, simulation results, model results and measurement results of the produced antenna were compared.