Application of radial basis function based neural networks to arrays of nonlinear antennas

This paper presents numerical examples that shows the results predicted by the RBF-NN models are consistent with those calculated from existing studies. It should be noted that the RBF-NN model is inherently one type of the general regression. This makes it powerful in modeling and predicting such a nonlinear problem. With the use of neural networks, the complex numerical computation about arrays of nonlinear antennas can be replaced by a very simple algebraic operation as the neural networks are well trained. Although the training work of a RBF-NN model is usually time consuming, it can be completed in advance. This study is useful in the applications of antenna design, remote sensing and wireless communication