Search academic texts

Information × Registration Number 0216U001198, 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 28-01-2016 Organization Ternopol National Economic University popup.description2 Object of research - the identification process of interdisciplinary mathematical models of complex objects under uncertainty based inductive approach. Purpose - to create theoretical foundations of interdisciplinary mathematical models (MM) changes in the state of complex objects (CO) in the life cycle and development of computational methods of implementation in real time; new means of mathematical modeling, based on the receipt of multiple estimates of parameters (state variables) CO and MM with desired properties predictors of real-time forecasting errors that are found within the measurement errors simulated characteristics. Research methods - methods of interval data analysis, methods of correlation analysis methods of the theory of identification, optimization techniques, methods of signal theory, methods, systems theory, methods of computer modeling. Results: The generalized approaches to building interdisciplinary MM in different fields such as medicine, the environment, energy, software engineering using an inductive approach; Mismatch between set specific prognostic providing model properties in traditional approaches and opportunities to ensure the accuracy of this because of errors in experimental data; proposed and proved the general principles and concept of construction of this class of models, based on an analysis of interval data, formulated the basic postulate of the concept of "theoretical achievements properties prognostic MM directly limited magnitude of errors in experimental data." Within the concept of an approach to resolve the contradiction between complexity and prognostic achievements properties: MM complexity must be such that the prediction error of MM was comparable to the level of accuracy of experimental data. The concept for the basic mathematical models of processes and changes in the state reasonably complex objects using discrete models of dynamical systems and mathematical models in the form of CO difference schemes. Proposed and justified criteria for evaluating the quality and complexity of the MM processes and complex objects in the form of difference schemes for building their case on the basis of inductive approach. Created new methods of structural identification of MM in the form of difference schemes based on "intelligence bee" (bee colony algorithm behavior). Proven benefits of the proposed structural identification methods compared to methods based on genetic algorithms. Improved methods of parametric identification of MM in a reasonable difference schemes with potential prognostic features and achievements relevant to their computational complexity of implementation. Specifically formulated mathematical problem of parametric identification of models in the form of interval systems of nonlinear algebraic equations and advanced search methods proposed solution of using these procedures adaptation step search. A number of new mathematical models in various fields such as medicine, the environment, energy, software engineering. A number of new software modules as tools to implement processes of identification interdisciplinary MM CO under uncertainty with specified prognostic characteristics and complexity of computer controlled implementation, which created the opportunity to solve a number of important practical problems related to decision-making based on different MM fields such as medicine, the environment, energy, software engineering. Scientific novelty is the first time within the inductive approach proposed and proved the general principles and concept of building class interdisciplinary models of complex objects and processes, based on an analysis of interval data and proven postulate that the theoretical achievements properties prognostic MM directly limited size of experimental error data collectively provide solution of one of the major contradictions in the mathematical modeling between complexity and prognostic properties achievements MM, based on the concept formulated generalized approaches to building interdisciplinary MM in different fields such as medicine, the environment, energy, software engineering using inductive approach; created new methods of structural identification of MM in the form of difference schemes based on "intelligence bee" (algorithms behavior of bee colonies), which provided a lower computational complexity of implementation compared to methods based on genetic algorithms; established methods of parametric identification of MM in the form of difference schemes, which are due to separation set all interval data on the principal and test parts according to the principles GMDH and consistent use of these parts in problem identification, marked by low computational complexity of implementation, which ensured reducing computational complexity and less than linear increase in computational complexity in the current iteration as the number of interval data; proposed and reasonably quantitative indicator to assess the computational complexity of implementation methods of parametric identification of MM in the form of difference schemes with random search of its parameters, which provided an opportunity to compare different computing implement these methods, regardless of the characteristics of computational tools. The results of research were tested at international and national conferences. Product Description popup.authors Брушніцька Анастасія Сергіївна Вальчишин Степан Петрович Голубєв Владислав Віталійович Гураль Ірина Володимирівна Крепич Світлана Ярославівна Мадюдя Ірина Анатоліївна Олійник Ірина Степанівна Пасічник Наталія Романівна Пасічник Роман Мирославович Порплиця Наталія Петрівна Протасевич Павло Тарасович Пукас Андрій Васильович popup.nrat_date 2020-04-02 Close