Condition Monitoring of Wind Turbines Through Image Similarity Evaluation

The condition monitoring of wind turbines is essential in a wide range of industrial fields for operational safety and efficiency. A smart visual system is proposed in this paper to achieve blade speed and tower vibration measurements through image similarity evaluation and advanced signal processing techniques. First, the similarity levels of a wind turbine image sequence are determined through the zero-normalized crosscorrelation algorithm. Then, an improved short-time autocorrelation method is utilized to process the resulting image similarity signal to derive the blade speed. Meanwhile, vibration information of the wind turbine tower is obtained through the spectral analysis of the image similarity signal of the tower region. The experimental data collected on a direct drive permanent magnet wind turbine illustrate that the blade speed and the tower vibration frequency can be measured accurately, confirming the feasibility of this technique for condition monitoring of wind turbines.