Computing And Analysis On Wind Turbine Generator Condition Monitoring Data Based On Spark Platform

Wind power in recent years to develop new renewable energy.Is an effective reduction of power generation costs and reduce environmental pollution of new energy sources.Can be a good integration into the national network to use.Ensure the safe,stable and efficient operation of wind turbine is the most important.The large-scale equipment condition monitoring of wind turbines has become an important part of wind power research.Power equipment condition monitoring will produce a lot of power data,data storage,data mining analysis is an important research content.With the popular Spark data processing platform to deal with the power of data,you can analyze the normal operation of electrical equipment,fault diagnosis,fault prediction has great advantages.The study of this paper is based on the Spark wind turbine monitoring data processing analysis of fault and early warning.In this paper,the research background of electric power equipment condition monitoring is introduced,and the difficulty of data processing technology of massive electric power at present is analyzed.Then the common fault of wind turbine is introduced.The wind power generation unit condition monitoring data has the characteristics of large power,data source,heterogeneous,complex and rapid growth.The existing fault diagnosis and early warning method is difficult to deal with large data in the case of precision to ensure rapid processing.In this paper,the research background,the wind power monitoring data mining.This paper introduces the cloud computing technology,Hadoop,HDFS,Spark and other related technologies and applications,focusing on the Spark framework for the operation of new models,distributed data sets(RDD),and gives its advantages for the later research Provide technical support advantages.Wind power data,multi-source,high-dimensional characteristics.In this paper,the wind power data are cleaned first,and the similarity measure between data is analyzed by stochastic forest method.The principal component analysis(PCA)is used to extract and normalize the data.And the algorithm of K-means is optimized.The Spark-based algorithm is applied to detect and process the wind noise data,which reflects the advantage of Spark wind power data preprocessing.Finally,the fault diagnosis model based on PCA and BP algorithm is constructed,and BP neural network model is constructed by training samples.Finally,the fault diagnosis of wind power based on Spark PCA and BP neural network is realized.The analysis of wind farm data processing fault diagnosis based on Spark has a good effect.