ALGORITHM OF PARAMETRIC IDENTIFICATION OF WELLS CONNECTIVITY SYSTEM

Abstract

Interaction analysis of injection and production wells is necessary for successful development of oil fields. Mathematically this problem can be reduced to the identification problem. In this article the problem of parametric multidimensional dynamic system identification of injection and production wells connectivity system has been considered. Existing parameter identification methods have been considered. The method of generalized adjusted object of measurement (GAOM) for solution of multidimensional dynamic system identification problem has been proposed. Existing identification methods (GAOM, prediction error method (Pem), subspace method of identification (N4sid)) have been implemented in Matlab. These algorithms have been tested for parametric identification of multidimensional dynamic system with zero initial conditions, with non-zero initial conditions, in the absence and presence of noise in measurement and control channels. The article contains comparative analysis of investigated methods which shows that GAOM is the most effective method for solution of the problem in the presence of noise in measurement and control channels. The bootstrap analysis has been applied to the proposed method, and this allows improving quality of identification. The research has revealed that GAOM using bootstrap analysis is superior to GAOM without using bootstrap analysis in identification quality. It is planned to use GAOM for parametric identification of wells connectivity system and to develop the algorithm for identifying non-stationary dynamical systems.