In order to solve the problem that the condition monitoring model of a single wind turbine is directly applied to the condition monitoring of other wind turbines in the wind farm with low accuracy,a cross-wind turbine parameter fine-tuning intelligent migration status monitoring method was proposed.The correlation between wind turbines in wind farm was calculated,and the representative normal wind turbine was selected.The input feature was constructed by using the massive normal SC AD A system data of representative wind turbine,and the condition monitoring model of wind turbine based on BiLSTM NN was established.The condition monitoring model is fine-tuned by using a large number of historical normal data of other wind turbines,and the personalized characteristic of wind turbine to be monitored was integrated to realize the intelligent migration condition monitoring across wind turbine.Using the real wind turbine data of a wind farm of cooperative enterprise,the proposed method was verified,and the result shows that the parameter fine-tuning migration condition monitoring method can detect early fault 2 months in advance compared with the single wind turbine condition monitoring model that directly monitors across wind turbine.Compared with other deep learning methods,the proposed method has higher prediction accuracy.
维普
万方数据