Analysing and Forecasting Degradation in Wind Turbines under Transient Operating Conditions through Vibration Analysis

In the field of wind turbines, there is growing attention towards. monitoring key components that are susceptible to high failure rates, such as gearbox, shafts, bearings, rotor blades, generators, etc. The usage of vibration sensors on turbine components aid in diagnosing and preventing breakdowns, ensuring reliable and efficient operation. A thorough understanding of degradation minimizes opportunity costs, optimizes maintenance expenses, and enables accurate prediction, planning, and effective mitigation of failures. In this study, two wind turbines from the same wind farm were considered for a detailed investigation of their vibration signature during normal operation. The vibrations were measured using identical sensors placed in the same locations (in both turbines) over an extended period, capturing a wide range of operating conditions. A multitude of methods including time domain analysis, frequency domain analysis, order analysis, envelope analysis were utilized to investigate and obtain a comprehensive understanding of the vibration dataset. These analyses helped identify the presence and extent of faults and abnormalities in the turbines. Moreover, the fault frequencies’ peaks derived from envelope analysis were cross-validated with analytically obtained fault frequencies. This research involves creating a distinctive degradation index, allowing the examination of degradation based on vibration behaviour over time. This effort facilitated the exploration of real-time changes in the degradation index, aiding in the ongoing assessment of wind turbine conditions. The analysis reveals multiple fault frequencies and greater degradation in the second turbine when compared to the first. Another key emphasis in this study involved the utilization of diverse autoregressive models, incorporating additional features to forecast the degradation index for the upcoming 15-day window. The resulting forecasts provide a clear tracking of degradation, offering advance notice for operators to implement timely predictive maintenance measures.