Research On Intelligent Electricity Information Collection And Analysis System Based On Big Data

With the rapid development of the intelligent distribution network,the collection dimension and frequency of user power consumption information data have increased significantly.Power user power consumption information constitutes a large amount of big data.How to analyze and process these massive electricity consumption information to obtain useful information is a question worthy of further study.Big data technology is used to perform data mining on the operation data of electrical equipment,and analyze the user's electricity consumption habits and equipment operation rules,which can help users improve their electricity consumption plans,and the power supply department can fully grasp the electricity consumption of users.Researching the intelligent data collection and analysis system based on big data can improve the electricity consumption level of enterprises and residents,reduce the cost of enterprises and the living expenses of residents,and has very important practical significance for the development of a low-carbon economy,energy conservation and emission reduction,and environmental protection.To this end,this article conducts research on the intelligent data analysis system based on big data.The specific work is as follows.Firstly,an intelligent power consumption information analysis system architecture based on Hadoop big data ecological cluster was proposed,and a parallel computing network composed of three physical machines was built.Intelligent power consumption information analysis system is composed of distributed computing framework Spark,Map Reduce,distributed storage mechanism HDFS,distributed management YARN,and distributed service framework Zookeeper,which realizes data collection,data storage,data management,and data of power consumption information.analysis.Secondly,using big data technology and user portraits to perform cluster analysis on user electricity usage rules,using the distributed parallel K-means algorithm under Mlib,and cluster analysis of smart electricity information data under the Spark platform framework.A numerical example shows that the algorithm has high calculation efficiency.Thirdly,a parallel DeepFM short-term load forecasting algorithm that takes into account the higher-order features of the load data is proposed.The high-order features and low-order features are imported into the Deep model and the FM model,respectively,and the training results are merged and imported into the full link layer.Output prediction results.Calculation examples show that the load prediction error can be reduced by introducing higher-order features of loaddata.The comparison with Gini coefficients and AUC curves of LR and FM algorithms proves the effectiveness of the algorithm.In this paper,a big data user power information analysis system is constructed.Using big data Hadoop and Spark technology to realize data collection,data analysis and data processing of user power characteristics.The parallel DeepFM short-term load prediction algorithm is used to improve the accuracy of short-term load prediction.It has great practical application value and wide application prospect.