Abnormal Detection of Wind Turbine Operating Conditions Based on State Curves

Abstract

Wind energy is a clean and renewable energy source and thus has promising future prospects. To increase the utilization rate and power generation of wind turbines and reduce their maintenance costs, it is necessary to monitor the operating conditions of wind turbines. This paper introduces a monitoring method based on state curves and includes a study that analyzes five types of state curves, namely, wind speed–power, wind speed–rotor speed, wind speed–pitch angle, rotor speed–power, and rotor speed–pitch angle. The results indicate that due to the external environment (e.g., atmospheric temperature, atmospheric pressure, wind turbulence, wind direction, topography), the wind turbine internal hardware performance, and the control strategy, the first three curves did not allow the wind turbine to distinguish the normal operation status from a fault status. However, the rotor speed–power and rotor speed–pitch angle curves were able to accurately monitor abnormal conditions of the wind turbine. This study aims to establish theoretical curves of a wind turbine and correct the state curves to calculate the distance from the actual operating point to the state curves. During operation, the wind turbine will be under several conditions, ranging from maximum power point tracking to constant speed to constant power conditions. Using the time window and a confusion matrix to determine the best deviation of the wind turbine under different operating conditions is more effective in reducing the false alarm rate. Based on the optimal offset distances and the relationship among rotor speed, power, and pitch angle during start-up and shut down, a corresponding evaluation system is established. Taking the operational data from a wind farm as an example, the research reveals that when a wind turbine is operating normally, the deviations in the state curves under different operating conditions fall within an appropriate range; once an abnormal condition occurs, if the number of abnormalities exceeds a specified value during a specified time window, then an alarm signal will sound. Compared with the supervisory control and data acquisition (SCADA) system, in this system the alarm time is advanced.