Global Performance Estimation Based on Gaussian Mixture Model for Wind Turbines

In order to ensure that the wind turbines are reliable in stable condition and economical in maintenance cost, the most effective way is to estimate and monitor the performance and operation of the wind turbine. Traditional fault diagnosis methods using multivariate statistical process usually assume the unit only has a single operating condition, so it’s not suitable for multi-regimes. Aiming at this problem, this paper proposed a global performance estimation method of multi-regimes condition based on Gaussian mixture model (GMM). First establish GMM to train the baseline model, cluster the sample data using the similar GMM method, and then calculate the distance between the baseline model and the GMM of sample data by two different methods. The result shows that this method can identify the characteristics of the turbine productivity well, and can detect the abnormality of power curve that is related to incipient fault.