INTEGRATION OF THE DYNAMIC STATE CLASSIFICATION MODEL INTO THE DIGITAL TWIN ARCHITECTURE OF COMPLEX TECHNICAL SYSTEMS.

Saved in:

Bibliographic Details
Title:	INTEGRATION OF THE DYNAMIC STATE CLASSIFICATION MODEL INTO THE DIGITAL TWIN ARCHITECTURE OF COMPLEX TECHNICAL SYSTEMS.
Alternate Title:	ІНТЕГРАЦІЯ МОДЕЛІ ДИНАМІЧНОЇ КЛАСИФІКАЦІЇ СТАНІВ В АРХІТЕКТУРУ ЦИФРОВОГО ДВІЙНИКА СКЛАДНИХ ТЕХНІЧНИХ СИСТЕМ.
Authors:	Vychuzhanin, V. V.¹ v.v.vychuzhanin@op.edu.ua, Vychuzhanin, А. V.¹
Source:	Informatics & Mathematical Methods in Simulation / Informatika ta Matematičnì Metodi v Modelûvannì. 2026, Vol. 16 Issue 3, p439-456. 18p.
Subjects:	Digital twin, Online data processing, Real-time computing, Engineering systems, Risk assessment
Abstract:	The article is devoted to solving the current scientific and technical problem of increasing the operational efficiency and metrological reliability of automated diagnostic systems for complex technical systems (CTS) under real-time operating conditions. The main focus is on overcoming the limitations of existing statistical models, which often do not account for the dynamics of degradation processes and the "concept drift" that occurs when the operating modes of equipment change. The propose and theoretically substantiate a methodology for integrating a modified dynamic state classification model (Discrete Cosine Transform - DCT) into the analytical loop of a digital twin (DT). The proposed approach is based on the formation of a closed cycle of data collection, preprocessing, and intelligent analysis, where the key diagnostic features are the parameters of structural (Rstr) and functional (Rfunc) risks. Unlike traditional classifiers operating in an offline batch mode, the proposed architecture is based on the stream processing paradigm, which allows for continuous adaptation of classification rules upon the arrival of each new sensor message. For the quantitative evaluation of the developed system's effectiveness, specific metrological indicators are introduced and tested: average forecast update delay ( Δupd ), stream computational efficiency (tproc), and forecast stability coefficient under input signal degradation (Rstab). Experimental verification, conducted on extended monitoring datasets of marine power plants and OREDA databases, showed that the integration of the prognostic model into the DT loop provides a significant increase in lead time - an average of 10–15 minutes before the occurrence of a critical event. This creates the necessary time margin for corrective measures by the operator or an automated control system. Special attention is paid to the intelligent error classifier module, which functions as a "selfhealing" loop of the diagnostic cycle. This module implements multi-level verification of classification results through dynamic physical simulation in the DT, allowing for effective localization of high-uncertainty zones in the risk space and minimizing the probability of false positives. The practical value of the results lies in the possibility of formalizing strict response regulations based on the verified Lead Time, which significantly reduces the risks of technogenic accidents. Future research prospects are related to scaling the model to network structures of interacting objects and implementing autonomous decision-making mechanisms without operator intervention, which will serve as the foundation for creating fully autonomous digital platforms for CTS lifecycle management. [ABSTRACT FROM AUTHOR]
	Copyright of Informatics & Mathematical Methods in Simulation / Informatika ta Matematičnì Metodi v Modelûvannì is the property of Odessa Polytechnic University and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	The article is devoted to solving the current scientific and technical problem of increasing the operational efficiency and metrological reliability of automated diagnostic systems for complex technical systems (CTS) under real-time operating conditions. The main focus is on overcoming the limitations of existing statistical models, which often do not account for the dynamics of degradation processes and the "concept drift" that occurs when the operating modes of equipment change. The propose and theoretically substantiate a methodology for integrating a modified dynamic state classification model (Discrete Cosine Transform - DCT) into the analytical loop of a digital twin (DT). The proposed approach is based on the formation of a closed cycle of data collection, preprocessing, and intelligent analysis, where the key diagnostic features are the parameters of structural (Rstr) and functional (Rfunc) risks. Unlike traditional classifiers operating in an offline batch mode, the proposed architecture is based on the stream processing paradigm, which allows for continuous adaptation of classification rules upon the arrival of each new sensor message. For the quantitative evaluation of the developed system's effectiveness, specific metrological indicators are introduced and tested: average forecast update delay ( Δupd ), stream computational efficiency (tproc), and forecast stability coefficient under input signal degradation (Rstab). Experimental verification, conducted on extended monitoring datasets of marine power plants and OREDA databases, showed that the integration of the prognostic model into the DT loop provides a significant increase in lead time - an average of 10–15 minutes before the occurrence of a critical event. This creates the necessary time margin for corrective measures by the operator or an automated control system. Special attention is paid to the intelligent error classifier module, which functions as a "selfhealing" loop of the diagnostic cycle. This module implements multi-level verification of classification results through dynamic physical simulation in the DT, allowing for effective localization of high-uncertainty zones in the risk space and minimizing the probability of false positives. The practical value of the results lies in the possibility of formalizing strict response regulations based on the verified Lead Time, which significantly reduces the risks of technogenic accidents. Future research prospects are related to scaling the model to network structures of interacting objects and implementing autonomous decision-making mechanisms without operator intervention, which will serve as the foundation for creating fully autonomous digital platforms for CTS lifecycle management. [ABSTRACT FROM AUTHOR]
ISSN:	22235744
DOI:	10.15276/imms.v16.no3.439