Adaptive Estimation Algorithms and their Applications to Measurement Data Processing
A. Stepanov Oleg, V. Motorin Andrei · 2019
The paper presents a review of methods for adaptive estimation of signals with unknown structures and parameters of models that determine the properties of both the signal itself and the errors of its measurement. The main approaches to solution of adaptive estimation problems are considered. One of them is aimed at designing algorithms with a learning sample, and the other one relies on the Bayesian approach and aims at solving the joint problem of signal estimation and identification of models used to describe the signal and measurement errors in real time. An adaptive estimation algorithm based on the Bayesian approach is described in detail. The problems in which the proposed algorithm is used are discussed. Its effectiveness is illustrated by the example of gravity anomaly estimation on a moving vehicle under model uncertainty of the signal and measurement errors.