Feed-forward neural network-based algorithm for intelligent positioning of direction-finding crossings

This paper proposes a feed-forward neural network(FNN) based algorithm for intelligent positioning of angle of arrival(AOA) to address the problems of complex computation, slow positioning speed and low positioning accuracy of traditional cross-location direction finding methods. The algorithm obtains the position of a known target from the available data, calculates its angle of arrival at the observatory, combines them into data pairs, and inputs them into a constructed neural network to obtain the corresponding model. The network can then locate unknown targets. It is verified that the algorithm has reliable localization accuracy and higher robustness than conventional algorithms. The proposed method also shows that the training of the neural network on the basis of a large amount of data can greatly reduce the operation time and achieve real-time target localization.