Research On Classification Model Under Spark Platform And Its Application In Power Network Equipment

In the process of intelligent construction of power grid,the monitoring system of power grid is huge,the number of monitoring data is increasing exponentially,and the problem of data processing becomes difficult.The flow processing platform Storm platform can realize real-time processing of data,but the model of data classification needs to be trained in advance.This paper mainly studies the classification module of power equipment monitoring data based on Spark platform.This paper puts forward two kinds of classification model,the first classification model using three ratio method to transformer continuous data into discrete data,for the discrete data analysis,random forest classification model training;training is dissolved in using the standard test data set and the gas in the transformer oil data set the accuracy test.In order to reflect the advantages of Spark platform relative to other platforms,the model on Hadoop is selected for performance comparison,and all of them have excellent performance.Second classification model is introduced into Xgboost algorithm using more of the Internet,and the principle of the algorithm is deduced,according to the principle of the algorithm is introduced to the transformer fault classification,as the classification model for training,and use standard data sets and the transformer oil dissolved gas data set on the classification accuracy of the model the test.The method of using PMML to transmit data between Spark and Storm is proposed.In this paper,a parallel transformer fault classification model combined with three ratios and random forests is proposed,and the Xgboost fault classification model is used after the three ratio.It is of great significance to select the classification model of data flow processing platform for Storm power equipment.