Research On Sampling Based Aggregate Query Method Of Power Quality Data

Power quality data are generated continuously with the operation of the power grid.Power quality data aggregation query is an important means to understand the power quality status of the power grid.Power quality data has the characteristics of a large amount of data and high dimension,and the acquisition of accurate query results often has a long delay,which cannot meet the requirements of interactive query in the power grid.Therefore,it may be an effective method to introduce the idea of the approximate query into a power quality data aggregation query and shorten query delay at the cost of sacrificing certain query accuracy.However,the existing sampling-based approximate query has some problems,such as large sampling error and delayed sample update,when facing the demand of power quality data aggregation query.To solve this problem,this paper proposes a sampling-based power quality big data aggregation query method.The main research contents are as follows:(1)When the natural stratified sampling method is applied to power quality big data aggregation query,the result has a large error.In order to solve this problem,a twostage stratified sampling method is proposed in this paper.Based on natural stratification,the clustering stratification method is used to further reduce the variance in the stratified sampling layer,to reduce the error of query results.The samples are stored in the memory in the form of multi-granularity sample sets,which can be quickly combined into logical samples when needed to meet the requirements of personalized query delay and reduce the sample storage space.Experimental results based on power quality data show that when the sampling rate is 5%,the accuracy of the proposed method can be improved by 12% ? 15% compared with the benchmark method.(2)The data of power quality conform to the characteristics of data flow,so the problem of sample updating is not timely.To solve this problem,a sample set maintenance method is proposed in this paper,which stratifies and samples the data items of real-time power quality data stream,and updates and maintains the multigranularity sample set with the latest data samples.Experimental results show that the sample set maintenance method can update the samples in time,keep the query precision of two-stage stratified sampling samples,and meet the real-time aggregation query requirements of business personnel.(3)Based on the above research content,this paper also implements the prototype system of power quality data aggregation query.The design and implementation of sample initialization,sample maintenance,and approximate query engine are introduced in detail,and the application effect of the prototype system is preliminarily demonstrated.