Search academic texts

Information × Registration Number 0215U000738, 0114U000569 , R & D reports Title Theory construction and methods for implementing a real-time interdisciplinary mathematical models of complex objects change state. popup.stage_title Head Dyvak Mykola Petrovych, Registration Date 07-04-2015 Organization Ternopol National Economic University popup.description2 Object of research are the processes of identification the interdisciplinary mathematical models of complex objects under uncertainty based on inductive approach. Purpose of research is to create theoretical bases for creation the interdisciplinary mathematical models (MM) of changing the states of complex objects (CO) in the life cycle and to develop the methods of their computational implementation in real-time; new tools for mathematical modeling based on obtaining multiple estimates of parameters (state variables) the CO and on MM with given predicting properties in real-time with predicting errors that are within the measurement errors of modeled characteristics. Research methods are the methods of interval data analysis, correlation analysis methods, methods of the identification theory, optimization techniques, methods of the signals theory and methods of systems theory, computer simulation methods. New results are: the generalized approaches to building interdisciplinary MM in different fields such as medicine, environment, energy, software engineering etc using an inductive approach are created; the mismatch was found between requirements of ensuring the certain predicting properties of model in traditional approaches and capabilities for ensuring this accuracy caused by the errors in the experimental data; the general principles and concept of construction these class of models, based on the analysis of interval data were proposed and proved; the basic postulate of the concept was formulated like "theoretically achievable predictive properties of MM directly limited by value of errors in experimental data"; within the concept an approach to solving the mismatch between the complexity and achievable predictive properties was proposed: MM complexity must be such that the prediction error of MM was comparable with errors in experimental data. In concept as basic mathematical model of changing the state of processes and complex objects using it is proved to use the discrete models of dynamical systems and mathematical models of CO in kind of difference schemes; the criteria for assessing the quality and complexity the structure of MM of processes and complex objects in the kind of difference schemes for case of their building on the basis of inductive approach were proposed and proved; the new methods of structure identification of MM in the kind of difference schemes based on "bee colony intellect" (algorithm of behavior the bee colony) was created; the advantage of the proposed structural identification methods compared to methods based on genetic algorithms was proved; the methods for parameter identification of MM in the kind of difference schemes with proved potentially achievable predictive features and relevant to their computational complexity of its implementation were improved; in particular the mathematical task of parameter identification of models in kind of interval systems of nonlinear algebraic equations was formulated and advanced methods of finding the solution of proposed systems using procedures with adaptation the search step were proposed. Product Description popup.authors Вовкодав Олександр Валерійович Гураль Ірина Володимирівна Крепич Світлана Ярославівна Манжула Володимир Іванович Олійник Ірина Степанівна Падлецька Наталія Ігорівна Пасічник Наталія Романівна Порплиця Наталія Петрівна Пукас Андрій Васильович Савка Надія Ярославівна Чайківська Юлія Мирославівна popup.nrat_date 2020-04-02 Close