Premium
Neural network‐based array synthesis in presence of obstacles
Author(s) -
Ayestarán Rafael G.,
LasHeras Fernando
Publication year - 2005
Publication title -
international journal of numerical modelling: electronic networks, devices and fields
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.249
H-Index - 30
eISSN - 1099-1204
pISSN - 0894-3370
DOI - 10.1002/jnm.595
Subject(s) - coupling (piping) , antenna (radio) , antenna array , artificial neural network , topology (electrical circuits) , computer science , set (abstract data type) , point (geometry) , obstacle , phase (matter) , field (mathematics) , process (computing) , radiation pattern , acoustics , electronic engineering , algorithm , physics , mathematics , telecommunications , engineering , electrical engineering , artificial intelligence , geometry , mechanical engineering , quantum mechanics , pure mathematics , law , political science , programming language , operating system
Traditional antenna array synthesis methods are based on the reconstruction of the array factor coefficients or an equivalent current distribution over a surface enclosing the antenna. The coefficients are then used to estimate the amplitude and phase of the feeding voltages that must be applied to the ports of the array in order to generate a radiated field distribution according to a given set of specifications. These approaches consider implicitly that the radiating elements are ideal, without any interaction between them. Coupling effects between the individual radiating elements once inserted in the array might modify their radiation properties and the necessary feeding values for a specified radiated field distribution, affecting significantly to the global behaviour of the array. Neural networks (NN) offer an efficient way to incorporate the real radiating properties and the coupling effects between the elements of an array in the synthesis process without increasing the complexity of the model from the designer's point of view. Once the coupling effects are modelled, they can be used to perform synthesis tasks in presence of any obstacle in a near environment, which can be modelled as a part of the array, with the induced currents established by the coupling effects taken into account in the NN learning procedure. Copyright © 2005 John Wiley & Sons, Ltd.