Sustainable Evaluation Of The Covariation Matrix Of An Object Noise Based On Iterative Processes For The Solution Of Nonlinear Operational Equations

Abstract

Algorithms for the stable estimation of the covariance noise matrix of a controlled object are
presented. To find the covariance noise matrix of an object, the possibilities of iterative
processes for solving nonlinear operator equations without the regularity property are analyzed.
To regularize the procedure for constructing the desired solution, a modification of the GaussNewton method is used, adapted to solve incorrect nonlinear problems. The above algorithms
make it possible to perform a stable estimation of the covariance matrix of the noise of an object
and thereby increase the accuracy of the adaptive estimation procedure.