Begell House Inc.
Journal of Automation and Information Sciences
JAI(S)
1064-2315
46
1
2014
On Using the Shcur Method for Solving Unilateral Quadratic Matrix Equation
1-8
Vladimir B.
Larin
S.P. Timoshenko Institute of Mechanics of National Academy of Sciences of Ukraine, Kyiv, Ukraine
The algorithm of solving the unilateral quadratic matrix equation is proposed. The algorithm is based on the Cayley transform of a matrix pencil and Schur algorithm, which is commonly used to find the solution of the algebraic Riccati equation. This algorithm allows one to find solution in the case of a singular matrix as well which is a coefficient of the higher degree of the equation. As shown in the example, the proposed algorithm is also effective in the case where all the matrix coefficients of this equation are degenerate. The algorithm efficiency is demonstrated in the examples that other authors have examined.
Hybrid Algorithm for Identification of Linear by States Hammerstein Model
9-19
Fedor G.
Garashchenko
Kyiv National Taras Shevchenko University, Kyiv, Ukraine
Olga G.
Moroz
Kiev National Taras Shevchenko University
The hybrid algorithm is proposed for identification of nonlinear dynamic Hammerstein model where static nonlinearity is modeled with radial basis functions neural network (RBFNN) and linear dynamic part − with a state-space model. The algorithm uses a particle swarm optimization (PSO) for RBFNN parameters estimating and subspace identification (SI) − for estimating the linear part parameters. Numerical example demonstrates the effectiveness of the PSO/SI algorithm proposed.
Guaranteed Estimation of Parameters of Linear Algebraic Equations for Nonstationary Observations
20-29
Alexander G.
Nakonechnyi
Kiev National Taras Shevchenko University, Kiev
Sergey V.
Demydenko
Kiev National Taras Shevchenko University
Estimates of solutions of linear algebraic equations with degenerated matrixes and unknown right-hand parts are considered. For nonstationary observations of equations solutions with errors, which are random processes with the given first and second moments, we investigate the problems
of representation of root-mean-square estimates using solutions of systems of algebraic equations.
Structural Recognition of the Nonlinear Discrete Dynamic Objects Based on the Generalized Probabilistic Criteria
30-41
Pavel A.
Kucherenko
Rostov State University of Railway Communication, Rostov-on-Don, Russia
Sergey V.
Sokolov
Rostov State University of Railway Transport (Russia)
The urgency of the research concerning the structural recognition problems of discrete multi-structured nonlinear objects observed in the presence of non-Gaussian interferences is proved. An approach to solving the problem on structure recognition of the discrete objects with a random structure change based on the generalized probabilistic criteria is offered. A model example for recognizing structures of the nonlinear object with a rapidly changing structure is considered to illustrate the efficiency of the proposed approach.
On the Problem of In-Flight Geometric Calibration Using Unknown Landmarks
42-52
Alexander I.
Tkachenko
International Research and Training Center of Information Technologies and Systems of National Academy of Sciences of Ukraine and Ministry of Education and Science of Ukraine, Kyiv
This article is initiated by the publication of the proposed by D.V. Lebedev method of in-flight geometric calibration of the on-board opto-electronic complex of a spacecraft by observations of unknown landmarks. Adverse effect of disturbing factors on the accuracy of determining the mutual attitude of the imaging camera and star sensor in the case of the spacecraft by images of the mentioned landmarks are studied. The means to weaken this effect are proposed.
Equations for the Survival Probability of an Insurance Company in the (B, S)-market Taking into Account Advertising
53-62
Boris V.
Bondarev
Donetsk National University, Ukraine
Valeria O.
Boldyreva
Donetsk National University
The activities of an insurance company, operating in the (B, S)-market, when a risky assets are described by the Samuelson model, and time interval varies from 0 do +∞, are considered. The sufficient conditions for the existence of derivatives of a survival probability as a function of the initial capital, are obtained. The integro-differential partial equations for the survival probability of an insurance company are derived.
Investigation of the Applicability of the Co-occurrence Matrix for Detecting Steganoaudiosignals
63-72
Natalya V.
Koshkina
V.M. Glushkov Institute of Cybernetics of National Academy of Sciences of Ukraine, Kiev
The sensitivity of rows of the co-occurrence matrix of audio signals computed in the time and the wavelet coefficients domains to distributed and sequential steganographic embedding by LSB method has been investigated. The relations between the steganographic capacity and the accuracy of SVM classification based on rows features of co-occurrence matrices stego and normal audios have been found. The method for determining the length of hidden messages by weighted voting of binary classifiers set that distinguish between two stego audios with different steganographic capacity has been proposed.
Matrix Integro-Differential Riccati Equation for Parabolic System
73-83
Miroslav M.
Kopets
National Technical University of Ukraine
"Igor Sikorsky Kiev Polytechnic Institute",
Kiev
The problem of minimization of quadratic functional on solutions of a system of linear parabolic equations is under consideration. Controls enter simultaneously both right-hand parts of the equations and boundary conditions. For investigation of the stated optimization problem the method of Lagrange multipliers is used. Such approach makes it possible to obtain necessary conditions of optimality. Under these conditions the matrix integro-differential Riccati equation with partial derivatives was derived.