Radial Basis Function Neural Network With Hidden Node Interconnection Scheme for Thinned Array Modeling

To extend the modeling area with artificial neural networks (ANNs) from finite periodic arrays to thinned arrays, where spacings between adjacent elements are crucial for array performance, an efficient model is proposed in this letter. Considering the spacings, a novel hidden node interconnection-radial basis function neural network (HNI–RBFNN) is developed to map the relationship between the array electromagnetic (EM) responses and the element ones. The element EM responses are obtained with the traditional RBFNN only involving the element geometry, while the connected weights of hidden layer nodes are determined by mutual coupling and array environment with the HNI scheme. A numerical example of the thinned phased array is used to evaluate the validity of the proposed model.