Research On Operation State Monitoring And Fault Diagnosis Of Wind Turbine Based On Data Mining

As a kind of clean energy,wind power has an signficant influence on the energy structure in China.With the rapid development of wind power industry,the maintenance of the wind turbine is getting more and more important.Due to the complex structure of wind turbine and the severe operating environment,it’s been a challenging and meaningful hot area for condition monitoring and fault diagnosis of wind turbine.A method based on improved DBSCAN clustering algorithm is proposed to achieve the condition monitoring and fault diagnosis;the Adaboost algorithm is combined with BP neural network to improve the performance of the wind turbine fault modeling.This paper has the following structrue.Starting from a brief overview of the data mining technology,the paper gives the main framework for it,then introduces the transmission mode of supervisory control and data acquisition(SCADA).With regard to the imcomplete and abnormal characteristics for the data collected from SCADA system,a preprocessing method cleaning,normalizing and feacture extracting from the raw data is proposed.Secondly,the improved DBSCAN clustering algorithm is used to monitor and identify the running state of the wind turbine.In view of the deficiency of the classical DBSCAN algorithm,the idea of piecewise fitting by distance distribution between data points is used to automatically achieve the radius,then we calculate the expectation of the size of points in the radius of the cluster.Then,the performance of the improved algorithm is verified by the historical data from SCADA system and classical data sets.Finally,the combination of Adaboost algorithm and BP neural network is adopted,BP neural network is used as weak predictor in Adaboost.We take the gearbox oil temperature as the target,the characteristic parameters related to the gear oil temperature are selected as input values.By analyzing and calculating the residual between the predicted value and the actual value,the fault warning and alarm threshold value is obtained,and the fault diagnosis of the wind turbine is realized.