Parametric Modeling of Unequally Spaced Linear Array Based on Artificial Neural Network

In this paper, a novel parametric modeling approach of electromagnetic (EM) behaviors of an unequally spaced linear array based on artificial neural network (ANN) is proposed. The ANN model takes the locations of array elements as input, with training and testing sample set obtained by full-wave EM simulations. After training, the ANN model can achieve the active element patterns (AEPs) of array elements rapidly once the locations of elements are input. Thus, the array radiation pattern including mutual coupling and platform effects can be obtained by superposition of AEPs. The validity and efficiency of the proposed approach are confirmed with a five-element unequally spaced linear array. The numerical experiment show that the mean absolute percentage error (MAPE) of the predicted pattern of the array is less than 5%. The proposed approach can be used in the real-time EM responses of the pattern synthesis of arrays.