Condition monitoring for multiple wind turbines based on balanced distribution adaptive transfer learning
Accurate condition monitoring of wind turbines is crucial to the safe and stable operation of wind turbines and the improvement of economic benefits.However,affected by the divergence in the distribution of operating data of different wind turbines,the existing condition monitoring methods have the problem of difficulty in taking into account the accuracy and efficiency in the application scenario of multiple wind turbines.BDA can shorten the data distance and reduce the data distribution divergence.Therefore,this paper propose a multi-wind turbine condition monitoring method based on balanced distribution adaptive transfer learning.Firstly,the mutual information method based on Copula entropy is used to mine the key influencing parameters of the wind turbine condition;then,a wind turbine condition monitoring model is established based on the GRU model and SPRT method;wind turbine operation data distribution assimilation model based on BDA is constructed,and used for multi-wind turbine condition monitoring.Results show that the proposed method can effectively save the modeling cost and calculation cost,and can significantly improve the monitoring efficiency on the premise of ensuring the monitoring accuracy of the operating state of multiple wind turbines.
wind turbinecondition monitorbalanced distribution adaptive transfer learningsequential probability ratio testgated recurrent unit
万方数据
维普
