THE NEURAL-NETWORK AND STATISTICAL ALGORITHMS FOR ESTIMATING COORDINATES OF A SOURCE OF RADIO RADIATION IN MULTI-POSITION RADIO SYSTEMS IN THE PRESENCE OF ABNORMAL ERRORS OF PRIMARY PARAMETER MEASUREMENT

摘要

The paper considers algorithms for estimating the coordinates of a radio radiation source by a single measurement in the angle-measuring and difference-type range-measuring multi-position radio systems in the presence of abnormal errors of measurement of primary parameters. The algorithms are developed based on neural networks with direct coupling. Analysis of accuracy characteristics of the neural-network estimation of coordinates is performed — as compared with the statistical Bayes-type estimation obtained with the use of Huber's minimax approach.