Simulation of Behavior and Intelligence

Optimal segmentation of structural experimental curves by dynamic programming

I. A. Kogan

Moscow

Abstract: The article is concerned with classifying the process state from indirect data represented as experimental structural curves. Structural curves described by stochastic difference autoregression equations are segmented in an optimal way by a dynamic programming algorithm. The autoregression parameter vector is assumed to take on at every time one of a finite set of values depending on the state of the curve source. For a source which may be in one of a finite set of states making a Markov chain a method is proposed of identifying a range of curves for whose segmentation the algorithm cam be employed in real time.


Received: 08.04.1987

 English version: Automation and Remote Control, 1988, 49:7, 934–942

Bibliographic databases:

•