Integrated approach to diagnosing complex technical systems: experimental validation and multidimensional efficiency assessment
DOI:
https://doi.org/10.33216/1998-7927-2025-291-5-5-17Keywords:
predictive diagnostics, Bayesian networks, CBR adaptation, failure simulation modeling, risk-oriented metrics, diagnostic stability, intelligent maintenanceAbstract
This paper presents a comprehensive experimental validation of an integrated approach to the diagnosis of the technical condition (TC) of complex technical systems (CTS), using ship power plants (SPPs) as an example. The proposed methodology combines precedent-based logic (Case Based Reasoning – CBR), probabilistic forecasting using Bayesian networks and Markov chains, and simulation modeling of degradation scenarios and cascading failures. Testing was conducted under three scenarios: normal operating mode, high-load mode, and a scenario with limited data availability, which enabled a thorough assessment of the algorithms' adaptability and resilience to changing operational factors. Classical binary classification metrics (Accuracy, Precision, Recall, and F1 score) were used for quantitative evaluation of diagnostic quality, along with newly introduced extended indicators: weighted accuracy (WAcc), F1 score accounting for the criticality of component failures (F1W), recall weighted by failure risk (RecallR), cost-adjusted precision for false alarms (PrecisionC), and the Diagnostic Stability Index (DSI). The results of the multi-scenario experiment showed a consistent improvement in all major indicators: Accuracy increased from 78.5% to 85.3%, Precision from 75.2% to 83.1%, Recall from 80.1% to 87.6%, F1 score from 77.5% to 85.3%, RecallR reached 91.0%, and DSI was 0.983. Five-fold cross-validation yielded a standard deviation of F1 score at 2.2%, confirming the reproducibility and reliability of the proposed method for experimental testing of the integrated diagnostic approach for CTS. The implementation of a cyclic procedure "simulation, probabilities, CBR adaptation" significantly reduced the number of false alarms and missed critical failures in SPP equipment. The practical significance of the approach lies in its potential integration into SCADA/PMS systems of marine CTS and ground power stations, facilitating a shift to intelligent predictive maintenance, thereby reducing unplanned downtime, lowering costs, and enhancing equipment reliability. Future research prospects include increasing the adaptability of the approach, expanding the precedent base, and developing tools for automated processing of heterogeneous data.
References
1. Vychuzhanin V., Vychuzhanin A.Stochastic Models and Methods for Diagnostics, Assessment, and Prediction of the Technical Condition of Complex Critical Systems / V. Vychuzhanin, A. Vychuzhanin. Kyiv :LihaPres, 2025. 360 p.https://doi.org/10.36059/978-966-397-457-6
2. Vychuzhanin V. V., Rudnichenko N. D.Metodyinformatsionnykhtekhnologiy v diagnostikesostoyaniyaslozhnykhtekhnicheskikhsistem: monografiya / V. V. Vychuzhanin, N. D. Rudnichenko. Odessa :Ekologiya, 2019. 178 p.
3. Yan H., Wang F., Yan G., He D.Hybrid approach integrating case-based reasoning and Bayesian network for operational adjustment in industrial flotation process / H. Yan, F. Wang, G. Yan, D. He // Journal of Process Control. 2021. Vol. 103. Pp. 34–47.https://doi.org/10.1016/j.jprocont.2021.05.003
4. Xu X., Lin Y., Ye C.Fault diagnosis of marine machinery via an intelligent data-driven framework / X. Xu, Y. Lin, C. Ye // Ocean Engineering. 2023. Vol. 289, No. 1. Article 116302.https://doi.org/10.1016/j.oceaneng.2023.116302
5. Schweidtmann A. M., Zhang D., von Stosch M.A review and perspective on hybrid modeling methodologies / A. M. Schweidtmann, D. Zhang, M. von Stosch // Digital Chemical Engineering. 2024. Vol. 10. Article 100136.https://doi.org/10.1016/j.dche.2023.100136
6. Nikpour H., Aamodt A.Fault diagnosis under uncertain situations within a Bayesian knowledge-intensive CBR system / H. Nikpour, A. Aamodt // Progress in Artificial Intelligence. 2021. Vol. 10, No. 1. Pp. 245–258.https://doi.org/10.1007/s13748-020-00227-x
7. Chen M., Xia J., Huang R., Fang W.Case-based reasoning system for aeroengine fault diagnosis enhanced with attitudinal Choquet integral / M. Chen, J. Xia, R. Huang, W. Fang // Applied Sciences. 2022. Vol. 12, No. 11. Article 5696.https://doi.org/10.3390/app12115696
8. Soleimani M., Campean F., Neagu D.Integration of Hidden Markov modelling and Bayesian network for fault detection and prediction of complex engineered systems / M. Soleimani, F. Campean, D. Neagu // Reliability Engineering & System Safety. 2021. Vol. 215. Article 107808.https://doi.org/10.1016/j.ress.2021.107808
9. El-Awady A., Ponnambalam K.Integration of simulation and Markov chains to support Bayesian networks for probabilistic failure analysis of complex systems / A. El-Awady, K. Ponnambalam // Reliability Engineering & System Safety. 2021. Vol. 211. Article 107511.https://doi.org/10.1016/j.ress.2021.107511
10. Zhong D., Xia Z., Zhu Y., Duan J.Overview of predictive maintenance based on digital twin technology / D. Zhong, Z. Xia, Y. Zhu, J. Duan // Heliyon. 2023. Vol. 9, No. 4. Article e14534.https://doi.org/10.1016/j.heliyon.2023.e14534
11. Kalafatelis A. S., Nomikos N., Giannopoulos A., Trakadas P.Survey on predictive maintenance in the maritime industry using machine and federated learning / A. S. Kalafatelis, N. Nomikos, A. Giannopoulos, P. Trakadas // TechRxiv. 2025. Vol. 11.https://doi.org/10.36227/techrxiv.173473250.04784922/v1
12. Özaydın E., Fışkın R., Uğurlu Ö., Wang J.A hybrid model for marine accident analysis based on Bayesian network (BN) and association rule mining (ARM) / E. Özaydın, R. Fışkın, Ö. Uğurlu, J. Wang // Ocean Engineering. 2022. Vol. 247. Article 110705.https://doi.org/10.1016/j.oceaneng.2022.110705
13. Cheliotis M., Lazakis I., Cheliotis A.Bayesian and machine learning-based fault detection and diagnostics for marine applications / M. Cheliotis, I. Lazakis, A. Cheliotis // Ships and Offshore Structures. 2022. Vol. 17, No. 12. Pp. 2686–2698.https://doi.org/10.1080/17445302.2021.2012015
14. OREDA.Offshore Reliability Data Handbook. 6th ed. OREDA, 2015.
15. Soliman A., Marsha M. A., Rahaman S.Digital twin predictive maintenance systems / A. Soliman, M. A. Marsha, S. Rahaman // Journal of Ocean Technology. 2024. Vol. 19, No. 2. Pp. 46–76.
16. Jovanović I., Karatuğ Ç., Perčić M., et al.Combined fault tree analysis and Bayesian network for reliability assessment of marine internal combustion engine / I. Jovanović, Ç. Karatuğ, M. Perčić, et al. // Journal of Marine Science and Application. 2025.https://doi.org/10.1007/s11804-025-00692-7
17. Velasco-Gallego C., Lazakis I.Mar-RUL: A remaining useful life prediction approach for fault prognostics of marine machinery / C. Velasco-Gallego, I. Lazakis // Applied Ocean Research. 2023. Vol. 140. Article 103735.https://doi.org/10.1016/j.apor.2023.103735
18. Schultheis A.Exploring a hybrid case-based reasoning approach for time series adaptation in predictive maintenance / A. Schultheis // ICCBR ’24 Workshop Proceedings. CEUR WS, Vol. 3708, 2024.
19. Daya A. A., Lazakis I.Developing an advanced reliability analysis framework for marine systems operations and maintenance / A. A. Daya, I. Lazakis // Ocean Engineering. 2023. Vol. 272. Article 113766.https://doi.org/10.1016/j.oceaneng.2023.113766
20. Neupane D., Bouadjenek M. R., Dazeley R., Aryal S.Data-driven machinery fault diagnosis: A comprehensive review / D. Neupane, M. R. Bouadjenek, R. Dazeley, S. Aryal // Neurocomputing. 2025. Vol. 627. Article 129588.https://doi.org/10.1016/j.neucom.2025.129588
21. Lv Y., Yang X., Li Y., Liu J., Li S.Fault detection and diagnosis of marine diesel engines: A systematic review / Y. Lv, X. Yang, Y. Li, J. Liu, S. Li // Ocean Engineering. 2024. Vol. 294. Article 116798.https://doi.org/10.1016/j.oceaneng.2024.116798
22. Li Y., Zhang W., Cui L., Gao H.System reliability modeling and analysis for a marine power equipment operating in a discrete-time dynamic environment / Y. Li, W. Zhang, L. Cui, H. Gao // Quality and Reliability Engineering International. 2024. Vol. 40. Pp. 3422–3438.
