[Système hybride HMM-RBFNN utilisant des caractéristiques pertinentes et non redondantes pour la classification des défauts de roulements]
This paper presents an innovative hybrid approach that combines Hidden Markov Models (HMM) with Radial Basis Function Neural Networks (RBFNN) for the automatic classification of mechanical faults in rolling element bearings using vibration signal analysis. The signals are sourced from the well-established database (https://engineering.case.edu/bearingdatacenter/welcome), widely used in fault diagnosis research. Raw signals are preprocessed to extract relevant features across time, frequency, and time–frequency domains, including wavelet packet decomposition. To enhance classification robustness and reduce computational complexity, dimensionality reduction is performed using Principal Component Analysis (PCA), complemented by Fisher score-based feature selection. HMMs are trained to capture the temporal dynamics of the signals, while RBFNNs leverage the reduced feature space for fine-grained classification. A comprehensive performance comparison is conducted between standalone HMM and RBFNN models, as well as their integration within the hybrid HMM-RBFNN system. Experimental results demonstrate that the proposed hybrid method significantly improves classification accuracy, highlighting its potential for industrial predictive maintenance applications.
Cet article présente une approche hybride innovante combinant les modèles de Markov cachés (HMM) et les réseaux de neurones à fonction de base radiale (RBFNN) pour la classification automatique des défauts mécaniques des roulements à billes à partir de l’analyse des signaux vibratoires. Ces signaux proviennent de la base de données (https://engineering.case.edu/bearingdatacenter/welcome), largement utilisée dans la recherche en diagnostic de défauts. Les signaux bruts sont prétraités afin d’extraire les caractéristiques pertinentes dans les domaines temporel, fréquentiel et temps-fréquence, notamment par décomposition en paquets d’ondelettes. Pour améliorer la robustesse de la classification et réduire la complexité de calcul, une réduction de dimensionnalité est effectuée par analyse en composantes principales (ACP), complétée par une sélection de caractéristiques basée sur le score de Fisher. Les HMM sont entraînés à capturer la dynamique temporelle des signaux, tandis que les RBFNN exploitent l’espace de caractéristiques réduit pour une classification fine. Une comparaison de performances exhaustive est menée entre les modèles HMM et RBFNN autonomes, ainsi que leur intégration au sein du système hybride HMM-RBFNN. Les résultats expérimentaux démontrent que la méthode hybride proposée améliore significativement la précision de la classification, soulignant ainsi son potentiel pour les applications de maintenance prédictive industrielle.
Révisé le :
Accepté le :
Publié le :
Mots-clés : Roulement, Diagnostic, Score de Fisher, Modèle de Markov caché (MMC), Analyse en composantes principales (ACP), Réseau de neurones à fonction de base radiale (RBF), Transformée en paquets d’ondelettes (WPT)
Miloud Sedira 1 ; Ahmed Felkaoui 1
CC-BY 4.0
Miloud Sedira; Ahmed Felkaoui. A hybrid HMM-RBFNN system using relevant and non-redundant features for bearing fault classification. Comptes Rendus. Mécanique, Volume 354 (2026), pp. 365-395. doi: 10.5802/crmeca.350
@article{CRMECA_2026__354_G1_365_0,
author = {Miloud Sedira and Ahmed Felkaoui},
title = {A hybrid {HMM-RBFNN} system using relevant and non-redundant features for bearing fault classification},
journal = {Comptes Rendus. M\'ecanique},
pages = {365--395},
year = {2026},
publisher = {Acad\'emie des sciences, Paris},
volume = {354},
doi = {10.5802/crmeca.350},
language = {en},
}
TY - JOUR AU - Miloud Sedira AU - Ahmed Felkaoui TI - A hybrid HMM-RBFNN system using relevant and non-redundant features for bearing fault classification JO - Comptes Rendus. Mécanique PY - 2026 SP - 365 EP - 395 VL - 354 PB - Académie des sciences, Paris DO - 10.5802/crmeca.350 LA - en ID - CRMECA_2026__354_G1_365_0 ER -
%0 Journal Article %A Miloud Sedira %A Ahmed Felkaoui %T A hybrid HMM-RBFNN system using relevant and non-redundant features for bearing fault classification %J Comptes Rendus. Mécanique %D 2026 %P 365-395 %V 354 %I Académie des sciences, Paris %R 10.5802/crmeca.350 %G en %F CRMECA_2026__354_G1_365_0
[1] Semi-automated diagnosis of bearing faults based on a hidden Markov model of the vibration signals, Measurement, Volume 127 (2018), pp. 141-166 | DOI
[2] Fault diagnosis of ball bearings using continuous wavelet transform, Appl. Soft Comput., Volume 11 (2011) no. 2, pp. 2300-2312 | DOI
[3] A tutorial on hidden Markov models and selected applications in speech recognition, Proc. IEEE, Volume 77 (1989) no. 2, pp. 257-286 | DOI
[4] A novel rolling bearing fault diagnosis and severity analysis method, Appl. Sci., Volume 9 (2019) no. 11, 2356 | DOI
[5] Artificial neural networks and genetic algorithm for bearing fault detection, Soft. Comput., Volume 10 (2006), pp. 264-271 | DOI
[6] Fault diagnosis of rotating machines based on the EMD manifold, Mech. Syst. Signal Process., Volume 135 (2020), 106443 | DOI
[7] Accurate bearing remaining useful life prediction based on Weibull distribution and artificial neural network, Mech. Syst. Signal Process., Volume 56–57 (2015), pp. 150-172 | DOI
[8] Deep learning-based intelligent fault diagnosis methods toward rotating machinery, IEEE Access, Volume 8 (2020), pp. 9335-9346 | DOI
[9] Design of intelligent fault diagnostic system (FDS), Process Saf. Environ. Prot., Volume 85 (2007) no. 6, pp. 566-578 | DOI
[10] Incipient fault diagnosis of roller bearing using optimized wavelet transform based multi-speed vibration signatures, IEEE Access, Volume 5 (2017), pp. 19442-19456 | DOI
[11] New fault recognition method for rotary machinery based on information entropy and a probabilistic neural network, Sensors, Volume 18 (2018), 337 | DOI
[12] Development of EBP-Artificial neural network expert system for rolling element bearing fault diagnosis, J. Vib. Control, Volume 17 (2011) no. 8, pp. 1131-1148
[13] An online bearing fault diagnosis technique via improved demodulation spectrum analysis under variable speed conditions, IEEE Syst. J., Volume 14 (2020) no. 2, pp. 2323-2334 | DOI
[14] Intelligent bearing fault diagnosis method combining compressed data acquisition and deep learning, IEEE Trans. Instrum. Meas., Volume 67 (2018) no. 1, pp. 185-195 | DOI
[15] Fault diagnosis method of rolling bearing using principal component analysis and support vector machine, J. Mech. Sci. Technol., Volume 32 (2018), pp. 5079-5088 | DOI
[16] Fault diagnosis of rolling element bearing using time-domain features and neural networks, 2008 IEEE Region 10 and the Third International Conference on Industrial and Information Systems, Kharagpur, India (2008), pp. 1-6 | DOI
[17] Comparison between traditional neural networks and radial basis function networks, 2011 IEEE International Symposium on Industrial Electronics, Gdansk, Poland (2011), pp. 1194-1199 | DOI
[18] Fault diagnosis of rotating machinery based on MFES and D–S evidence theory, Proceedings of the 24th Chinese Control and Decision Conference (CCDC), Taiyuan (2012), pp. 1624-1629 | DOI
[19] Using radial basis function networks for function approximation and classification, ISRN Appl. Math., Volume 2012 (2012), 324194 | DOI | MR
[20] Bearing fault detection based on hybrid ensemble detector and empirical mode decomposition, Mech. Syst. Signal Process., Volume 41 (2013) no. 1–2, pp. 510-525 | DOI
[21] A machine learning approach for the condition monitoring of rotating machinery, J. Mech. Sci. Technol., Volume 28 (2014) no. 1, pp. 61-71 | DOI
[22] et al. A wavelet packet spectral subtraction and convolutional neural network based method for diagnosis of system health, J Mech. Sci. Technol., Volume 33 (2019), pp. 5683-5687 | DOI
[23] Comparison between hidden Markov models and artificial neural networks in the classification of bearing defects, Rotating Machinery and Signal Processing. SIGPROMD’2017 (A. Felkaoui; F. Chaari; M. Haddar, eds.) (Applied Condition Monitoring), Volume 12, Springer, Cham, 2017 | DOI
[24] Intelligent fault diagnosis of rolling bearing and gear system under fluctuating load conditions using image processing technique, J. Mech. Sci. Technol., Volume 34 (2020), pp. 4107-4115 | DOI
[25] Artificial intelligence for fault diagnosis of rotating machinery: a review, Mech. Syst. Signal Process., Volume 108 (2018), pp. 33-47 | DOI
[26] Rolling bearing fault convolutional neural network diagnosis method based on casing signal, J. Mech. Sci. Technol., Volume 34 (2020) no. 6, pp. 2307-2316 | DOI
[27] Classification of ball bearing faults using a hybrid intelligent model, Appl. Soft Comput., Volume 57 (2017), pp. 427-435 | DOI
[28] A hybrid generalized hidden Markov model-based condition monitoring approach for rolling bearings, Sensors, Volume 17 (2017) no. 5, 1143 | DOI
[29] Fault diagnosis and prognosis of bearing based on hidden Markov model with multi-features, Appl. Math. Nonlinear Sci., Volume 5 (2020) no. 1, pp. 71-84 (Sciendo) | DOI | MR
[30] Review rolling element bearing diagnostics using the Case Western Reserve University data: a benchmark study, Mech. Syst. Signal Process., Volume 64–65 (2015), pp. 100-131 | DOI
[31] A novel rolling bearing fault diagnosis method based on adaptive feature selection and clustering, IEEE Access, Volume 9 (2021), pp. 99756-99767 | DOI
[32] A review of gear fault diagnosis using various condition indicators, Procedia Eng., Volume 144 (2016), pp. 253-263 | DOI
[33] Online bearing clearance monitoring based on an accurate vibration analysis, Energies, Volume 13 (2020), 389 | DOI
[34] Prognostics and health management design for rotary machinery systems—reviews, methodology and applications, Mech. Syst. Signal Process., Volume 42 (2014), pp. 314-334 | DOI
[35] Hidden Markov Models based intelligent health assessment and fault diagnosis of rolling element bearings, PLoS One, Volume 19 (2024) no. 2, e0297513 | DOI
[36] Rolling element bearing diagnostics—a tutorial, Mech. Syst. Signal Process., Volume 25 (2011) no. 2, pp. 485-520 | DOI
[37] Intelligent bearing fault diagnosis method combining mixed input and hybrid CNN–MLP model, Mech. Syst. Signal Process., Volume 180 (2022), 109454 | DOI
[38] Ball bearing vibration data for detecting and quantifying spall faults, Data Brief., Volume 47 (2023), 109019 (PMID: 36942099; PMCID: PMC10023968) | DOI
[39] A novel hybrid deep learning method for fault diagnosis of rotating machinery based on extended WDCNN and long short-term memory, Sensors, Volume 21 (2021) no. 19, 6614 | DOI
[40] Fault diagnosis of rolling bearings using PCA-based hybrid neural networks, Measurement, Volume 189 (2022), 110462 | DOI
[41] Attention-based CWT–CNN–BiLSTM framework for intelligent bearing fault diagnosis, Mech. Syst. Signal Process., Volume 195 (2025), 110893 | DOI
[42] An optimized hybrid IHHO–DBN–ELM model for rolling bearing fault diagnosis, Expert Syst. Appl., Volume 236 (2025), 121353 | DOI
[43] Hybrid diagnostic framework for interpretable bearing fault classification using CNN and dual-stage feature selection, Sensors, Volume 25 (2025) no. 20, 6386 | DOI
[44] Bearing fault diagnosis method based on deep convolutional neural network and random forest ensemble learning, Sensors, Volume 19 (2019) no. 5, 1088 | DOI
Cité par Sources :
Commentaires - Politique