We present the results of an efficient analysis–based on adaptive network fuzzy inference systems (ANFIS)–of the impedance an infinite array of rectangular microstrip antennas. In the proposed modeling approach, first, a number of input–output pairs (inputs include spacing between antennas, outputs include resonant frequency, quality factor, and input resistance) are computed from full-wave methods, and they are then used in the training process. An advantage of the proposed model in comparison with full-wave and approximate methods is that, after convergence of the training process, it bypasses the repeated use of complex computations for new inputs presented to it without any restriction. In addition, a comparison of the proposed model with those related to a single rectangular microstrip antenna shows considerable differences, especially for input resistance and quality factor, which is of importance in practical applications.
