IDENTIFICATION OF DEPENDENCES ON THE BASIS OF AUTOREGRESSIVE MODELS
A.N.Tyrsin
Chelyabinsk
Pages: 40-46 Subsection: IMAGE AND SIGNAL ANALYSIS
Abstract
A new method of identification of dependences on the basis of autoregressive models is proposed. The distinctive feature of the method is verification of the degree of closeness of dimensionless autoregression coefficients to tolerance ranges of the models considered rather than comparison of selected equations. The agreement between trend and autoregressive models is demonstrated. Operation of the method is illustrated by an example of identifying the distance covered by a car after the stop signal as a function of speed.