Communication Dans Un Congrès Année : 2025

A Divisive Unsupervised Feature Selection Approach for Explainable Remaining Useful Life Prediction

Résumé

Predicting the Remaining Useful Life (RUL) in maintenance often encounters challenges such as high dimensionality, feature redundancy, and limited explainability. This paper presents a novel approach that combines Interpretable Divisive Feature Clustering (IDFC) with Long Short-Term Memory (LSTM) networks. The IDFC algorithm leverages the strengths of variable clustering methods (VARCLUS) and the Clustering of Variables around Latent Components (CLV) to identify significant features and non-orthogonal latent components. This method enables effective dimensionality reduction by selecting key features rather than combining them. Integrating IDFC with a single-layer LSTM and Shapley Additive Explanations (SHAP) results in a robust and interpretable framework for RUL prediction, achieving a balance between accuracy and transparency. Experimental results on a bearing dataset show that the IDFC + LSTM model outperforms traditional methods while enhancing interpretability through the identification of key energyrelated features which influence the RUL prediction more.

Fichier principal
Vignette du fichier
dexa_119.pdf (1.23 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-05229009 , version 1 (08-09-2025)

Licence

Identifiants

Citer

Mouhamadou Lamine Ndao, Genane Youness, Ndèye Niang, Gilbert Saporta. A Divisive Unsupervised Feature Selection Approach for Explainable Remaining Useful Life Prediction. Database and Expert Systems Applications. DEXA 2025, Aug 2025, Bangkok, Thailand. pp.270-286, ⟨10.1007/978-3-032-02088-8_19⟩. ⟨hal-05229009⟩
461 Consultations
191 Téléchargements

Altmetric

Partager

  • More