Neural net aided fault diagnostics of large antenna arrays

Faulty elements identification in (large) antenna arrays from measured radiation fields/intensities is a problem of considerable theoretical and practical relevance. Given a set of N (complex) feeding currents and radiator effective heights, the state of the array is represented by a point in {0,1}/sup N/, where "0" and "1" denote the faulty and working state, respectively. Diagnostics reduces to the minimization among all possible array states of a functional, which measures the distance between the measured data and the field-intensities produced at the same point by the unknown state of the array. Standard minimization methods do not always yield the sought solution, due to possible trapping in local (spurious) minima. In this paper we present an asynchronous, discrete implementation of Vidyasagar\´s (see IEEE Trans. on Automatic Control, vol. AC-40, p.1359-75, 1995) mean field neural net based algorithm featuring excellent comparative performance both in terms of reliability and speed.