Innovative Strategies for Wind Turbine Reliability: Leveraging Vibration Data for Early Prognosis

Within the scope of this research work, a unique approach to estimating the state of high-speed bearings in wind turbines is presented. The method makes use of raw data that has been gathered from sensors that measure vibrations. In order to classify bearings into two distinct categories, the technique requires the calculation of a Kurtosis Spectrogram for each of the fifty samples and the utilization of a 1D-CNN. These bearings are classified as either Medium-life Expectancy (more than 15 days) or Short-life Expectancy (less than 15 days). The study underlines the significance of early identification in the process of estimating the Remaining-Useful Life (RUL) of machine components. This is essential for preventing failures and making maintenance decisions that are optimal. Based on recall, the 1D-CNN, which was built using Keras, has achieved an impressively high level of accuracy, reaching an overall accuracy of approximately 90%. The purpose of this study is to highlight the potential application of this technology in alert systems. It provides essential support for making technical decisions on component maintenance, which in turn ensures the dependability and efficiency of wind turbine systems.