CARLE: A Hybrid Deep-Shallow Learning Framework for Robust and Explainable RUL Estimation of Rolling Element Bearings

TL;DR

CARLE is a hybrid deep-shallow learning framework that improves robustness and explainability in RUL estimation of bearings by combining advanced neural networks with ensemble methods and thorough interpretability analysis.

Contribution

This paper introduces CARLE, a novel hybrid AI framework integrating deep and shallow learning for accurate, robust, and explainable RUL prediction of rolling element bearings under varying conditions.

Findings

CARLE outperforms state-of-the-art methods in accuracy.

The framework demonstrates high robustness under noise and domain shifts.

Interpretability analysis confirms transparency of the model.

Abstract

Prognostic Health Management (PHM) systems monitor and predict equipment health. A key task is Remaining Useful Life (RUL) estimation, which predicts how long a component, such as a rolling element bearing, will operate before failure. Many RUL methods exist but often lack generalizability and robustness under changing operating conditions. This paper introduces CARLE, a hybrid AI framework that combines deep and shallow learning to address these challenges. CARLE uses Res-CNN and Res-LSTM blocks with multi-head attention and residual connections to capture spatial and temporal degradation patterns, and a Random Forest Regressor (RFR) for stable, accurate RUL prediction. A compact preprocessing pipeline applies Gaussian filtering for noise reduction and Continuous Wavelet Transform (CWT) for time-frequency feature extraction. We evaluate CARLE on the XJTU-SY and PRONOSTIA bearing datasets. Ablation studies measure each component's contribution, while noise and cross-domain experiments test robustness and generalization. Comparative results show CARLE outperforms several state-of-the-art methods, especially under dynamic conditions. Finally, we analyze model interpretability with LIME and SHAP to assess transparency and trustworthiness.