Нейросетевой метод оценки угловых координат радиолокационных целей в цифровой антенной решетке

Abstract

A neural network method for measuring angular coordinates is proposed, which consists in combining a neural network with a diagram-forming scheme of a digital antenna array and forming a signal corresponding to the angular position of the target at the output of the neural network. A theorem is formulated that the output of the first layer of the proposed neural network generates signals corresponding to the beam patterns of the receiving channels of the digital antenna array. Using the example of a linear digital antenna array, the monopulse method for measuring angular coordinates is compared with the proposed neural network method. It is established that the training of the neural network should be carried out with the addition of noise, and training examples should be formed within the sector of angles occupied by the rays of the measuring channels. It is shown that the proposed method is able to compete with the monopulse method in the width of the angular sector, in which it is possible to estimate the angular coordinates and allows measuring the angular coordinates for non-identical beams of the monopulse group.