Actuality of the problem of parametric identification of a mathematical model

- Куропятник, Данило Іванович (2018) Actuality of the problem of parametric identification of a mathematical model Computer Science & Software Engineering : Proceedings of the 1st Student Workshop (CS&SE@SW 2018), Kryvyi Rih, Ukraine, November 30, 2018 (2292). pp. 70-75. ISSN 1613-0073

[img]
Preview
Text
paper06.pdf - Published Version

Download (650kB) | Preview

Abstract

The purpose of the article is to study the possibilities of increasing the efficiency of a mathematical model by identifying the parameters of an object. A key factor for parametrization can be called the consideration of properties of the values of the model at a specific time point, which allows a deeper analysis of data dependencies and correlation between them. However, such a technique does not always work, because in advance it is impossible to predict that the parameters can be substantially optimized. In addition, it is necessary to take into account the fact that minimization reduces the values of parameters without taking into account their real physical properties. The correctness of the final values will be based on dynamically selected parameters, which allows you to modify the terms of use of the system in real time. In the development process, the values of experimentally obtained data with the model are compared, which allows you to understand the accuracy of minimization. When choosing the most relevant parameters, various minimization functions are used, which provides an opportunity to cover a wide range of theoretical initial situations. Verification of the correctness of the decision is carried out with the help of a quality function, which can identify the accuracy and correctness of the optimized parameters. It is possible to choose different types of functional quality, depending on the characteristics of the initial data. The presence of such tools during parametrization allows for varied analysis of the model, testing it on various algorithms, data volumes and conditions of guaranteed convergence of functional methods.

Item Type: Article
Keywords: mathematical model, machine learning, optimization, parametrization, quality functional
Subjects: Science and knowledge. Organization. Computer science. Information. Documentation. Librarianship. Institutions. Publications > 00 Prolegomena. Fundamentals of knowledge and culture. Propaedeutics > 004 Computer science and technology. Computing. Data processing > 004.9 ІКТ ( Application-oriented computer-based techniques ) > 004.94 Simulation
Divisions: Institute for Digitalisation of Education > Joint laboratory with SIHE “Kryvyi Rih National University”
Depositing User: Сергій Олексійович Семеріков
Date Deposited: 08 Jan 2019 10:19
Last Modified: 23 Aug 2019 12:39
URI: https://lib.iitta.gov.ua/id/eprint/713305

Downloads

Downloads per month over past year

Actions (login required)

View Item View Item