Diagnosis of bearing faults in wind turbine systems using vibrational signal processing and machine learning

This study employs an integrated methodology for the analysis and diagnosis of bearing faults in rotating machinery and wind turbine systems.The methodology begins by analyzing the original signal using Variational Mode Decomposition to extract distinct modes.Subsequently, the envelope is derived from the optimal mode and transformed into the frequency domain using Fast Fourier Transform to compute the envelope spectrum.The spectrum is segmented into specific frequency bands, and the energy within each band is quantified as features for training a K Nearest Neighbors classification model.The dataset is partitioned into training and testing subsets using cross-validation, and model performance is assessed using metrics such as accuracy and F1 score to ensure robust diagnostic capabilities.Comparative analysis of frequency spectra from real wind turbine signals highlights improvements in energy localization and distribution post-envelope processing.The proposed methodology is then applied to classify faults using the Case Western Reserve University dataset, demonstrating significant enhancements in diagnostic accuracy.These findings underscore the efficacy of the methodology in advancing fault diagnosis in complex machinery systems.