Research On User-side Power Quality Based On Big Data Analysis Technology

The development of smart power grid is on the rise recently,many nonlinear elements has been applied to the power grids,which brings out negative influences to the quality of electric energy.In addition,China’s economy has developed rapidly,the demand for higher quality in power supply also grows in public.Therefore,conventional data analysis cannot satisfy the new demand confronting with the huge monitoring-data.So,it is an urgent affair for the power enterprise to meet the electricity demand by analyse the current electric circumstance.This paper has found ways to solve the predicament above by focusing on the solution in terms of voltage quality.The main contents are as follows:At first,Introducing the background and significance of power-quality.Comparing the current power-quality research method with the traditional one.And then,expounding the big data technology to verify the feasibility of using big data technology in power-quality analysis.Secondly,Analyzing the basic parameters of power-quality and tough issues in it.And then,to use data-mining algorithm to handle the power-quality data including k-medoids algorithm and k-GMM algorithm.The mathematical models of the two algorithms are built based on the measurement of cluster partition and realized by Java.At last,On the premise of analyzing the characteristics of power quality data,this paper use big-data-analysis method and Hadoop platform to solve the power-quality problems.Firstly,build a YARN-based Map Reduce framework and to process the power-quality data by modifying the internal parameters.Then,combined K-medoids algorithm and k-GMM algorithm on Hadoop platform to analyse the outlet voltage data of transformers.The result reveals that the methods created in this paper are reliable and valuable under the current standard of distribution-network.The methods raised in this paper can not only improve the existed power-quality analysis module,but also predict the trend of power quality data.In this experiment,two algorithms are verified the dependability and effectiveness in power-quality analysis.Also,Hadoop platform and Map Reduce framework has proved the outstanding efficiency and good accuracy in data-mining.The whole experiment proves the feasibility of using big-data technology in state evaluation analysis.