Intelligent monitoring and predictive maintenance of wind turbines: analyzing performance under ideal and real conditions

This thesis presents an intelligent monitoring and predictive maintenance framework for wind turbines, integrating Supervisory Control and Data Acquisition (SCADA) systems and advanced data analytics to bridge the gap between ideal and real-world performance. The increasing global demand for sustainable energy has accelerated the deployment of wind energy systems. However, operational inefficiencies, environmental variability, and mechanical failures continue to challenge optimal wind turbine performance. Using the FX EVO 30B-100 wind turbine as a case study, the research investigates the discrepancy between expected and actual energy production. This includes a detailed analysis of real-world power curves, wind resource estimations, and environmental factors such as turbulence intensity and air density. SCADA alarm data is systematically categorized and correlated with production losses to diagnose operational faults. The thesis compares SARIMA and LSTM models for short-term power and wind speed prediction. Results show that SARIMA performs well for stable short-term forecasts, while LSTM more effectively tracks real-world data trends. Additionally, a Random Forest classifier is trained to predict fault severity levels based on key SCADA parameters, offering an extra layer of decision support for maintenance scheduling.