Data-Driven Predictive Maintenance Framework for High-Speed Shaft Bearings Using Spectral Kurtosis and Adaptive Learning Approaches

Predicting the remaining useful life (RUL) of wind turbine bearings is crucial for ensuring reliability, minimizing unplanned downtime, and reducing maintenance costs in wind energy systems. However, existing prognostic models often suffer from limited accuracy due to their inability to effectively handle noisy vibration data and capture complex degradation patterns. This research presents an enhanced RUL prediction framework for high-speed bearing prognosis model, integrating data-driven signal processing with adaptive machine learning algorithms for improved fault progression analysis. The proposed approach leverages extracted time-frequency features and health indicators to train a hybrid regression model, enabling robust RUL estimation under varying operational conditions. Experimental validation using real-world bearing datasets demonstrates superior prediction accuracy and computational efficiency compared to baseline methods. The outcomes of this research can significantly improve predictive maintenance planning, enhance turbine reliability, and contribute to the sustainable operation of wind energy systems.