Research On Stator Temperature Data Cleaning Of Wind Power SCADA Based On Optimal Interclass Variance Algorithm

Research on stator temperature data cleaning of wind power SCADA based on Optimal Interclass Variance algorithm,through the analysis of wind farm SCADA system for the blower unit running state of generator running parameter acquisition,by data processing tools to store the parameters of the stator temperature related to the database.Inefficient because of the SCADA data analysis process,the model precision is not high,and the problem of insufficient economic benefit,need cleaning on their data sources and optimization of data pretreatment such as research,the accuracy of the quality of the original data and the model has obvious improvement.During the use and research of the cleaning algorithm of data preprocessing,the basic principle and use of neural network are combined to validate the data pretreatment method.The main work is as follows:(1)SCADA data processing research.Briefly introduced the structure of SCADA system and its function as data source.The pre-processing tools of SCADA data are used to prepare the data source to the research.The tool is to facilitate the use of data software independently developed independently in the early stage of data.The design framework of the data processing software,the interface generation script and the software description are emphasized,which provides the research guarantee and platform for the research of stator temperature in this paper.(2)The use of the data cleaning algorithm and explore.In order to improve the precision of the stator model,the related parameters of the quality of the data source of the stator temperature had absolute effect,through the analysis of the components of wind power SCADA system to collect data,for which the generator stator temperature,optimizing the process of data processing and analysis,here mentioned cleaning algorithm of two kinds of data preprocessing,fitting curve near the threshold method and field.The two algorithms are based on the most standard data points,the former fitting out the standard curve,the latter with the standard point as the circle.(3)OIV research.Need to rely on regular data cleaning algorithm combining with standard data points,the algorithm is proposed,the algorithm is mainly through slip is worth to the ideal threshold,and the threshold value in the whole data set,a place for abnormal,is normal,and normal part of the data set is the late need to modeling data source.The main feature of this algorithm is that it does not need to rely on standard data points in the process of data preprocessing.Compared with conventional two algorithms,it can be used directly without looking for standard points.(4)The OIV algorithm is improved.And the arithmetic structure of the original algorithm is improved.The original single threshold partition is changed to the limit of the double threshold,and the kernel point of the optimal group is the use of the double threshold value.The optimal formula through the example analysis shows that the improved algorithm is feasible and efficient,be able to deal with the temperature curve of the generator stator accurate data and by using neural network to forecast,a significant increase of wind turbine generator stator temperature prediction accuracy,the final threshold and fitting curve method,near field method and the optimal formula compare difference algorithm,the original on the precision of the model has been further improvement.