Research On Data Storage And Analysis Of Smart Grid State Monitoring Based On Cloud Computing

With the continuous development of research and construction of the smart grid,the level of intelligence and digitization of the power grid is getting higher and higher,and the depth and intensity of the condition monitoring on the power equipment of smart grid is also getting bigger and bigger.The condition monitoring data collected from power equipment grows exponentially,how to store massive state monitoring data effectively and analyze it effectively in order to assess the status of power equipment accurately has become a hot research issue.The traditional stand-alone environment is facing the problem of insufficient storage and computing resources,and can not satisfy the requirement of the state monitoring data processing.In this thesis,the cloud computing technology is introduced to the field of condition monitoring of smart grid.By introducing the distributed file system and the improvement of the traditional density clustering algorithm and the improvement of parallelization design,the storage and the partition of large state data is effectively solved.it provides a feasible method for the application of cloud computing in condition monitoring.This paper mainly do the following work:1)The problems in data processing of smart grid condition monitoring are analyzed,and the ideas and methods of data processing are studied;2)In view of the increasing scale of power equipment condition monitoring data,the traditional storage method can not solve the problem of large scale of condition monitoring data,a state monitoring data storage system based on cloud platform is designed.The distributed file system HDFS and Hbase database are used to storage status monitoring data,this storage system is prepared for the next step of the state monitoring data analysis and evaluation;3)Aiming at the problem of the shortcomings of the traditional density clustering algorithm,a clustering algorithm named DBCLustering based on density cluster structure is designed.The algorithm first constructs an index structure CR-Tree,which stores the core reachability of data nodes,then extracts a sorted linear table about the data reachability.Then the clustering results are output according to the result of the linear table.In order to solve the problem of computing capacity shortage of stand-alone algorithm,a Spark-based condition monitoring data clustering algorithm called RDD-DBClustering algorithm is proposed.The algorithm implements the parallelization of DBClustering algorithm under Spark platform,and it improves the ability of algorithm to deal with large-scale data.4)A Spark cluster consisting of nine nodes was built in the lab.The clustering algorithm of the RDD-DBClustering algorithm proposed in Chapter 5 was applied to the data of the insulator leakage current collected in the lab.The experimental results show that the parallel algorithm is better than the stand-alone version in the processing efficiency,and the algorithm has good parallelism,which is suitable for clustering analysis of the large state data.