Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents

Anatoly Baranovsky; Daniil Bobrikov; Igor Molodkin

doi:10.20295/2413-2527-2026-347-18-34

VAK Russia 1.2.1 UDC 004.051 VAK Russia 2.9.8

3 more Collapse

FORMATION OF SEPARATING SURFACES FOR CLASSES OF TECHNICAL STATES IN THE TASKS OF TRAINING INTELLIGENT DIAGNOSTIC AGENTS Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents

АА Аа

Published in Intellectual Technologies on Transport · Issue 3, 2026 · Pages 18–34 · Rubric: ARTIFICIAL INTELLIGENCE AND TRANSPORT SYSTEMS

DOI: https://doi.org/10.20295/2413-2527-2026-347-18-34 · EDN: JKBHEH

Received: 01.07.2026 Accepted: 26.08.2026 Published: 01.10.2026

Authors

Anatoly Baranovsky

Daniil Bobrikov

Igor Molodkin

Anatoly Baranovsky

Academic degree: candidate of technical sciences

Emperor Alexander I St. Petersburg State Transport University Associate Professor · “Information and Computing Systems” Department · Russian Federation · St. Petersburg

Daniil Bobrikov

Emperor Alexander I St. Petersburg State Transport University “Information and Computing Systems” Department · Russian Federation · St. Petersburg

Igor Molodkin

Emperor Alexander I St. Petersburg State Transport University Senior Lecturer · “Information and Computing Systems” Department · Russian Federation · St. Petersburg

ABSTRACT

The increasing complexity of information management systems (IMS) and the volume of telemetric data require a transition from traditional centralized diagnostic schemes to intelligent distributed systems that are adaptive and scalable. One of the key tasks in developing such solutions is to formalize the processes of forming regions and separating surfaces for technical states. Purpose: to develop a mathematical model and algorithms for forming regions and separating surfaces for technical states during the training of a prototype intelligent system capable of self-learning and predictive assessment of the technical condition of equipment under conditions of a large flow of diagnostic data. Methods: an approach is proposed that combines multi-agent organization of components, implementation of distributed streaming data processing using neural network and threshold methods, as well as the use of Big Data infrastructure for data storage, aggregation, and analysis. The diagnostic task has been formalized; algorithms for forming regions and separating surfaces of technical states, for predicting and assessing the probability of failure have been developed; a prototype has been implemented on a computing platform. Results: an architecture of an intelligent IMS has been developed, which includes agents for data collection, preprocessing, and analysis, a prediction module based on LSTM, a decision-making agent, and interfaces for interaction with the operator. Calculations are presented that demonstrate an increase in diagnostic reliability with an increase in data volume. Real-time telemetry processing scenarios have been implemented. Practical significance: the results obtained during the experiment indicate the effectiveness of the proposed approach. The possibilities for scaling the system and integrating it into the existing transport infrastructure are described, ensuring fault tolerance and autonomy.

KEYWORDS

multi-agent systems, neural network technologies, information management systems, transport, automation

Text

Text (RU) (PDF)

Read Download

1. Baranovsky A. M., Bobrikov D. A. Sovremennye podkhody k diagnostike i vosstanovleniyu rabotosposobnosti informatsionno-upravlyayushchikh sistem podvizhnogo sostava s primeneniem mnogoagentnykh i neyrosetevykh tekhnologiy [Modern Approaches to Diagnostics and Recovery of Rolling Stock Information Management Systems Using Multi-Agent and Neural Network Technologies], Intellektualnye tekhnologii na transporte [Intellectual Technologies on Transport], 2025, no. 1 (41), pp. 5–13. DOI:https://doi.org/10.20295/2413-2527-2025-141-5-13 (In Russian) EDN: https://elibrary.ru/IIIKAH

2. Baranovsky A. M., Bobrikov D. A. Mnogoagentnaya arkhitektura dlya intellektualnoy diagnostiki podvizhnogo sostava: kontseptsiya i prototipirovanie [Multi-Agent Architecture for Intelligent Rolling Stock Diagnostics: Concept and Prototyping], Intellektualnye tekhnologii na transporte [Intellectual Technologies on Transport], 2025, no. 3 (43), pp. 25–32. DOI:https://doi.org/10.20295/2413-2527-2025-343-25-32 (In Russian) EDN: https://elibrary.ru/UEUHQE

3. Baranovsky A. M., Musienko A. S. Dinamicheskie diagnosticheskie modeli i metod obespecheniya ustoychivosti kontrolya tekhnicheskogo sostoyaniya bortovykh sistem upravleniya letatelnykh apparatov [Dynamic Diagnostic Models and Method for Ensuring the Stability of Monitoring the Technical Condition of On-Board Control Systems of Aircraft], Trudy MAI, 2024, no. 139, 31 p. (In Russian) EDN: https://elibrary.ru/WVXLBK

4. Zvonarev S. V. Osnovy matematicheskogo modelirovaniya: uchebnoe posobie [Fundamentals of Mathematical Modeling: A Textbook]. Yekaterinburg, Ural Federal University Publishing House, 2019, 112 p. (In Russian) EDN: https://elibrary.ru/ZDEUBF

5. Romkin M. V. Identifikatsiya obektov zheleznodorozhnogo podvizhnogo sostava po ikh geometricheskim parametram [Identification of Objects of the Railway Vehicles Via Their Geometric Parameters], Identifikatsiya sistem i zadachi upravleniya (SICPRO ‘15): trudy X Mezhdunarodnoy konferentsii [System Identification and Control Problems (SICPRO ‘15): Proceedings of the X International Conference], Moscow, Russia, January 26–29, 2015. Moscow, V. A. Trapeznikov Institute of Control Sciences of RAS, 2015, pp. 738–750. (In Russian) EDN: https://elibrary.ru/TQGDTV

6. Maharramov Z. T., et al. Tekhnologiya Big Data: potentsial, problemy i primenenie v meditsine i zdravookhranenii [Big Data Technology: Potential, Challenges and Applications in Medicine and Healthcare], Radioelektronika i informatika, 2020, no. 1, pp. 54–68. (In Russian)

7. Karnaukhov N. S., Ilyukhin R. G. Vozmozhnosti tekhnologiy “Big Data” v meditsine [Capabilities of “Big Data” Technologies in Medicine], Vrach i informatsionnye tekhnologii [Medical Doctor and Information Technologies], 2019, no. 1, pp. 59–63. (In Russian) EDN: https://elibrary.ru/ZAFWGL

8. Yazinin V. S., Baranovskiy A. M., Vorobyev A. A., Romanova I. Yu. Ispolzovanie iskusstvennykh neyronnykh setey na Rossiyskikh zheleznykh dorogakh dlya kontrolya tokopriemnikov poezdov [The Use of Artificial Neural Networks on Russian Railways to Control Current Collectors of Trains], International Journal of Advanced Studies: Transport and Information Technologies, 2023, Vol. 13, no. 1, pp. 267–287. DOI:https://doi.org/10.12731/2227-930X-2023-13-1-267-287 (In Russian) EDN: https://elibrary.ru/UGLBCH

9. Tikhonov A. A. Bolshie dannye i glubokoe mashinnoe obuchenie v iskusstvennykh neyronnykh setyakh [Big Data and Deep Machine Learning in Artificial Neural Networks], Nauka i obrazovanie segodnya, 2018, no. 6 (29), pp. 35–38. (In Russian) EDN: https://elibrary.ru/XRHGYX

10. Rithani M., Kumar R. P., Doss S. A Review on Big Data Based on Deep Neural Network Approaches, Artificial Intelligence Review, 2023, vol. 56, iss. 12, pp. 14765–14801. DOI:https://doi.org/10.1007/s10462-023-10512-5 EDN: https://elibrary.ru/OVDDPQ

11. Axak N. G. Kontseptsiya postroeniya multiagentnykh sistem raspredelennoy neyrosetevoy obrabotki bolshikh dannykh [Concept of Construction of Multiagent Systems of Distributed Neural Network Processing of Big Data], Vestnik Khersonskogo natsionalnogo tekhnicheskogo universiteta [Visnyk of Kherson National Technical University], 2018, no. 3-1 (66), pp. 205–212. (In Russian)

12. Letunova I. E., Letunov E. D. Multiagentnaya sistema dlya analiza bolshikh dannykh na predpriyatiyakh raketnokosmicheskoy otrasli [Multi-Agent System for Analysis of Big Data at an Enterprise of the Rocket and Space Industry], Aktualnye problemy aviatsii i kosmonavtiki: Proceedings of the VIII International Scientific and Practical Conference, Krasnoyarsk, Russia, April 11–15, 2022. Vol. 2. Krasnoyarsk, Reshetnev Siberian State University of Science and Technology, 2022, pp. 45–47. (In Russian) EDN: https://elibrary.ru/WSNEHL

13. Lipatov A. G. Vozmozhnosti ispolzovaniya iskusstvennogo intellekta dlya upravleniya bolshimi informatsionnymi massivami dannykh Big Data [Possibilities of Using Artificial Intelligence to Manage Large Information Arrays of Big Data], Innovatsii i investitsii [Innovation and Investment], 2023, no. 5, pp. 187–189. (In Russian) EDN: https://elibrary.ru/UPHPKT

14. Bova V. V., Duckardt A. N. Primenenie iskusstvennykh neyronnykh setey dlya kollektivnogo resheniya intellektualnykh zadach [Application of Artificial Neural Networks for Collective Decision of Complex Intelligent Problems], Izvestiya Yuzhnogo federalnogo universiteta. Tekhnicheskie nauki [Izvestiya Southern Federal University. Engineering Sciences], 2012, no. 7, pp. 131–138. (In Russian) EDN: https://elibrary.ru/PAXXRB

15. Nikonov A. A. Gibridnye podkhody k adaptivnomu upravleniyu organizatsionnymi strukturami mnogoagentnykh sistem v usloviyakh nepolnoy nablyudaemosti i dinamichnykh izmeneniy sredy [Hybrid Approaches to Adaptive Management of Organizational Structures of Multi-Agent Systems Under Conditions of Incomplete Observability and Dynamic Changes in Environment], Vestnik Nauki, 2025, no. 1-2 (82), pp. 1022–1026. (In Russian) EDN: https://elibrary.ru/BNXBQM

PDF JATS XML

Editor-in-Chief

Anatoly Khomonenko

khomon@mail.ru

Deputy Editor-in-Chief

Lesya Bozhko

Principal Contact

Maksim Orzhevsky

206@it-pgups.ru

Editorial Staff

Staff Editorial

itt-pgups@yandex.ru +7(812)457-86-06

106

downloads

GOST

Baranovsky A. M., Bobrikov D. A., Molodkin I. A. Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents // Intellectual Technologies on Transport. 2026. no. 3. pp. 18-34. DOI: https://doi.org/10.20295/2413-2527-2026-347-18-34. EDN: JKBHEH. (Date of access 04.11.2026).

MLA

Anatoly Baranovsky, Bobrikov D. A., Molodkin I. A.. "Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents." Intellectual Technologies on Transport.3 (2026): 18-34. DOI: https://doi.org/10.20295/2413-2527-2026-347-18-34. EDN: JKBHEH.

APA

Baranovsky A. M., Bobrikov D. A., Molodkin I. A. (2026). Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents. Intellectual Technologies on Transport(3), 18-34. https://doi.org/10.20295/2413-2527-2026-347-18-34 EDN: JKBHEH.

Chicago

Anatoly Baranovsky, Bobrikov D. A., Molodkin I. A.. "Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents." Intellectual Technologies on Transport, no. 3 (2026): 18-34. https://doi.org/10.20295/2413-2527-2026-347-18-34. EDN: JKBHEH.

WIKI

{{cite journal | author = Baranovsky A. M., Bobrikov D. A., Molodkin I. A. | title = Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents | journal = Intellectual Technologies on Transport | year = 2026 | issue = 3 | pages = 18-34 | doi = 10.20295/2413-2527-2026-347-18-34 }} EDN: JKBHEH

Copy

Export: BibTeX EndNote RefMan RusMARC

FORMATION OF SEPARATING SURFACES FOR CLASSES OF TECHNICAL STATES IN THE TASKS OF TRAINING INTELLIGENT DIAGNOSTIC AGENTS Formation of Separating Surfaces for Classes of Technical States in the Tasks of Training Intelligent Diagnostic Agents

Confirmation