Begell House Inc.
Journal of Automation and Information Sciences
JAI(S)
1064-2315
42
11
2010
Parameters Identification by Gradient Methods of the Problems of Substance Diffusion in Nanoporous Medium
1-17
10.1615/JAutomatInfScien.v42.i11.10
Ivan V.
Sergienko
V.M. Glushkov Institute of Cybernetics of National Academy of Sciences of Ukraine, Kyiv, Ukraine
Vasiliy S.
Deineka
V. M. Glushkov Institute of Cybernetics of National Academy of Sciences of Ukraine, Kyiv, Ukraine
theory of optimal control
explicit expressions
gradients of functional-discrepancies
identification
gradient methods
parameters
problems of diffusion of substance
nanoporous medium
On the basis of theory of optimal control we construct explicit expressions for gradients of functional-discrepancies for identification by gradient methods of different parameters of problems of diffusion of substance in nanoporous medium.
Accuracy of Estimation of Parameters of Linear Regression on Errors in Variables
18-30
10.1615/JAutomatInfScien.v42.i11.20
Nikolay N.
Salnikov
Institute of Space Research of National Academy of Sciences of Ukraine and National Space Agensy of Ukraine, Kyiv, Ukraine
ultimate accuracy
estimation of parameters of linear regression
bounded errors
measurements of input variable
regressor
nonstochastic estimation methods
polyhedron method
numerical example
We consider ultimate accuracy of estimation of parameters of linear regression under the presence of bounded errors of measurements of input variable and regressors on usage of several main nonstochastic estimation methods. It was shown that for sufficiently low level of errors of regressors the minimax approach makes it possible to obtain exact values of estimated parameters. For high level of these errors only polyhedron method with explicit consideration of regressors errors makes it possible to obtain exact values of parameters on realization of usual requirements to sequences of errors and input data without errors. The obtained results are illustrated by means of the numerical example.
Reconstruction of the Initial State Vector of Nonlinear Automata over a Finite Ring
31-35
10.1615/JAutomatInfScien.v42.i11.30
Vladimir G.
Skobelev
Institute of Applied Mathematics and Mechanics, National Academy of Sciences of Ukraine, Donetsk, Ukraine
nonlinear automata
initial state vector reconstruction
problem complexity
parametric identification
reversible automaton
For Mealy and Moore automata over a finite commutative-associative ring with unity, which transition function is defined by nonlinear equations of the second degree, and the output function is affine and linear map of the states set, respectively, the problem of reconstruction of the initial state vector is solved. The case, when this problem is trivial, is considered. It is found that in other cases the problem is difficult. It is found that the property "to be reversible automaton" generally does not affect the complexity of solving the problem of reconstruction of the initial state vector for the investigated automata.
Comparative Analysis of Estimation Methods of Vertices Correlation while Bayesian Networks Construction
36-45
10.1615/JAutomatInfScien.v42.i11.40
Petr I.
Bidyuk
Institute for Applied System Analysis of National Technical University of Ukraine "Igor Sikorsky Kiev Polytechnic Institute", Kiev
Vladimir I.
Davidenko
Joint stock company 'Raiffeisen Bank Aval", Kiev, Ukraine
Dmitriy V.
Trofimenko
Educational-scientific complex Institute of Applied System Analysis of National Technical University of Ukraine "Kiev Polytechnic Institute", Ukraine
Alexander N.
Terentyev
Institute of Applied System Analysis of National Technical University of Ukraine "Kiev Polytechnical Institute", Kiev, Ukraine
Bayesian network construction
vertices correlation estimating
heuristic algorithm
information correlation metrics
metrics comparative analysis
While Bayesian networks construction the estimation methods for the correlation between the vertices are analyzed, using the heuristic algorithm. The theoretical justification of methods is carried out, the results of their practical use while construction of classical networks are considered and the algorithm of carrying out experiments on the basis of pseudo-random generation of Bayesian networks is described. The results, obtained for every method, were compared. The conclusions about the applicability of the considered methods of estimating the correlation between the vertices while constructing Bayesian networks are made.
Compression of Text Image Based on Selection of Characters and Their Classification
46-57
10.1615/JAutomatInfScien.v42.i11.50
Vladimir G.
Ivanov
National Yaroslav Mudry Academy of Law of Ukraine, Kharkov, Ukraine
Mikhail G.
Lyubarskiy
Yaroslav Mudryi National Academy of Law of Ukraine, Kharkov, Ukraine
Juriy V.
Lomonosov
National Yaroslav Mudry Academy of Law of Ukraine, Kharkov, Ukraine
text image compression
degree of compression
character image
classifying metric
The method for compression of text images based on new classifying metric the use of which enabled one to decrease essentially (more than twice) the number of classes of characters images is offered and investigated. This made possible to increase the degree of compression by 20 % on average as compared with currently the best algorithm of text images compression IB2 (DjVu).
Bayesian Recognition Procedures on Networks
58-63
10.1615/JAutomatInfScien.v42.i11.60
Alexandra A.
Vagis
V.M. Glushkov Institute of Cybernetics of National Academy of Sciences of Ukraine, Kiev, Ukraine
polynomial algorithm
Bayesian recognition procedure
network structure
learning sample
The polynomial algorithms of determining Bayesian network structure are described as "tree" or "polytree". In the known Bayesian network structure consideration is given to Bayesian recognition procedures built on learning samples by estimations of transition probabilities.
Magnetic Attitude Control and Stabilizing System of Egyptsat-1 Spacecraft
64-70
10.1615/JAutomatInfScien.v42.i11.70
Nikolay V.
Yefimenko
Research and Production Enterprise "Hartron-UCOM", Zaporozhye
attitude control and stabilizing system
electromagnets
three-component magnetometer
angular movement parameters
Results of the flight test carried out with magnetic attitude control and stabilizing system of the Egyptsat-1 spacecraft are considered. The system uses electromagnets as magnetorquers and three-component magnetometer as a sensor of source information on angular movement parameters. There is a brief description of control system software and algorithmic support and results of the flight test for the present system.
Method of Audio Signals Watermarking Based on the Wavelet and Fourier Transforms
71-80
10.1615/JAutomatInfScien.v42.i11.80
Natalya V.
Koshkina
V.M. Glushkov Institute of Cybernetics of National Academy of Sciences of Ukraine, Kiev
audio signals watermarking
wavelet and Fourier transforms
marking intangibility and robustness
new method and algorithm
The spectral approach to building systems with digital water marks (DWM), based on the characteristics of human hearing, and the analysis of effect of typical audio processing operations upon the audio spectrum, is considered. The new method and algorithm for permanent audio signals marking on the basis of fast wavelet transform and fast Fourier transform is proposed. The testing on intangibility and robustness of DWM embedded was carried out.