Application of Artificial Neural Network for Planar Printed Antenna Optimization

This paper presents the application of artificial neural network (ANN) in optimizing planar printed antennas Data generation for training the multilayer neural network derived from full wave based solver to improve the accuracy of network. Back propagation algorithm and momentum method is applied for updating the weights during each iteration. Nonlinear Sigmoid transfer function has been used for hidden neurons while a linear function is chosen for the output neurons. Interpolation technique has also been applied on the data set to reduce the computation time on the data generated by full wave based CAD software and to improve the accuracy by presenting large data. The paper also presents the comparison of test result with full wave based solver for validation and its computation time showing reduction.