Optimum design and production of seljuk star shaped microstrip antenna using artificial intelligence methods

With the developing technology and wireless communication systems being preferred frequently in recent years, microstrip antennas that are smaller, lighter and require performance are frequently preferred. Studies for different designs of these antennas are increasing in order to achieve the desired performance for the purpose of use. In the design of antennas, the calculation of the physical parameters that will provide the desired antenna characteristics such as radiant frequency input impedance and gain has become an important problem. In recent years, antennas designed with artificial intelligence methods based on the antenna design problem can achieve successful results. Meta-Heuristic Algorithms, which are frequently preferred in large-scale optimization problems, have become popular in scientific studies on microstrip antenna optimization. In this study, a new model for Neural Network training has been developed by combining back propagation algorithm and Meta-Heuristic algorithm. The biggest disadvantage of the back propagation algorithm in finding solutions is that it gets stuck at the local minimum rather than the global minimum. In this new hybrid training algorithm, local and global search are performed simultaneously. Initially, optimization algorithms were used to obtain the Neural Network weights, since the probability of catching the local minimum is low due to the long jump. Then, it is combined with Backpropagation algorithm to use local search capability in neural network training. Levenberg-Marquardt back propagation algorithm was used in the training phase of the Artificial Neural Network. In this study, Seljuk Star shaped microstrip antenna design, which can improve bandwidth compared to commonly used antennas, has been realized and simulated. The data set consisting of 1342 microstrip antennas obtained from the simulation results was trained and tested with the ANN model using 7 different optimization algorithms. In addition, antennas designed with materials with different dielectric coefficients were produced and the measurement results were obtained and compared with the simulation results.