Beam Pattern Optimization Method Based on Radial Basis Function Neural Network

In this paper, a method of beam pattern optimization based on Radial Basis Function Neural Network (RBFNN) is proposed for controlling sidelobe level of arbitrary geometry array. The proposed method takes advantage of the nonlinear mapping between the input and output of the radial basis function neural network, because of the nonlinear relationship between the position of the elements and the weighted vector of array in the Olen beamforming method. Many positions with errors centered on the real element positions are generated, when the beam pattern obtained by Olen beamforming method meet the design requirements, the corresponding positions and weighted vector are recorded as the input and output of training data. The beam patterns of uniform linear array, uniform arc array and random circular array are designed by using the trained neural networks. The results show that the proposed method is effective.