Abstract

This research study introduces a novel method for predicting the condition of high-speed bearings in wind turbines. The method utilises unprocessed data collected from sensors that measure vibrations. The process entails calculating a Kurtosis Spectrogram for each of the 50 samples and use a 1D-CNN to categorise bearings into two groups: Medium-life Expectancy (over 15 days) or Short-life Expectancy (under 15 days). This study highlights the importance of early detection in forecasting the Remaining-Useful Life (RUL) of machine components, which is essential for averting failures and enhancing maintenance decisions. The 1D-CNN, implemented using the Keras framework, has remarkable accuracy, achieving an overall accuracy of almost 90% as evaluated by recall. This study emphasises the potential application of this technology in alert systems, offering crucial assistance in making technical decisions about component upkeep, guaranteeing the dependability and effectiveness of wind turbine systems.