Bearing estimation using Hopfield neural network

Abstract

A neural network algorithm for bearing estimation is introduced. It utilizes a basic and proven property of Hopfield neural networks, i.e. the guaranteed convergence to a local minimum of the Lyapunov energy function. Unlike the previous methods, the new method estimates the in-phase and quadratic components separately and in a parallel manner and combines them to estimate the bearings of plane waves to an array. The connection parameters of the neural networks are calculated for both components with a significant reduction in computation in comparison with the previous methods. Furthermore, the new method is able to estimate the actual magnitude of each bearing component, rather than just its presence. This is accomplished by using the 1984 Hopfield model rather than the 1982 model, as opposed to the previous methods.<>