Compensation of Environmental and Operational Variations in Wind Turbine Natural Frequencies using SCADA data

Modal tracking is a fundamental tool that enables continuous dynamic Structural Health Monitoring (SHM). Yet, trends in identified natural frequencies, taken in isolation, cannot be uniquely attributed to structural changes – even less to occurring damage. In fact, modal parameters are influenced by environmental and operational variations (EOVs). These unde-sired influences are critical in the monitoring of all types of Civil Engineering structures, including wind turbines, where wind speed and direction govern operating state and loading, thereby altering the vibrational response. Here, a data-driven approach is presented to tackle this problem. Using regression models and Machine Learning (ML)-enhanced data analysis, the proposed methodology minimizes external effects using Supervisory Control and Data Acquisition (SCADA) measurements from a utility-scale onshore turbine. The portion of frequency variability attributable to am-bient and operating conditions is thus predicted and then subtracted to yield residuals linked primarily to mechanical condition. Control-chart analysis of the residuals demonstrates sub-stantially reduced EOV sensitivity and improved responsiveness to incipient faults. These results show that EOV-compensated modal tracking provides robust SHM indicators for wind turbines and increases sensitivity to damage-related anomalies.