Remaining Useful Life Prediction on Rolling Bearing Using Optimum Feature Extraction Integrated to Transformer Network

Predicting rolling bearings' Remaining Useful Life (RUL) is essential for predictive maintenance, but obtaining features that are crucial for model builders in industrial applications is extremely difficult. This study presents a novel approach that integrates optimum feature extraction with a Transformer network to enhance RUL estimation accuracy. A comprehensive feature selection process is applied to extract the most informative indicators from vibration signals, leveraging statistical, frequency-domain, and time-frequency domain features. These features are then fed into a Transformer-based deep learning model, which effectively captures long-term dependencies and complex temporal patterns in bearing degradation. Experimental validation using benchmark datasets demonstrates that the proposed method significantly outperforms conventional machine learning and deep learning models in terms of prediction accuracy and robustness. The findings highlight the potential of Transformer networks, combined with optimized feature extraction, as a powerful tool for intelligent prognostics in rotating machinery.