Research And Application Of Cloud Computing On Substation Status Monitoring System

With the improvement of substation informatization and digitalization,the amount of data generated by state monitoring of substation equipment increases exponentially.How to store and calculate these massive data reliably and evaluate the state of substation equipment in real time has become a research hotspot.Traditional single machine storage and computing capacity is limited,can not meet the needs of mass state monitoring data storage and computing.In recent years,with the rise of cloud computing technology.Internet companies have applied it in their own businesses(real-time recommendation)and made huge profits.In this paper,cloud computing technology is combined with substation state monitoring to solve the storage,offline calculation and online state evaluation problems of massive state monitoring data through distributed file system,data warehouse and real-time computing technology.The main research contents include:(1)Research mainstream cloud computing technology.The mainstream technologies in Hadoop ecosystem are studied,mainly including the design idea of distributed computing framework,system architecture,etc,to provide theoretical basis for the follow-up work.(2)According to the characteristics of the substation status monitoring system and the advantages of cloud computing technology,analyze the functional requirements of the system and design the overall architecture of the system,including data generation,data access,data processing and Web platform.The data warehouse scheme based on dimension modeling was proposed for the off-line computing platform,and the hierarchical structure and ETL processing process of the data warehouse were designed.(3)Spark Streaming,a distributed real-time computing framework,was used in substation state monitoring to analyze the application of Streaming data processing in state monitoring.The residual capacity of substation battery is predicted by BP neural network algorithm,and the trained model is saved as PMML model file,then the trained model is loaded by Spark Streaming to predict the residual capacity of battery in real time,and the battery status is evaluated according to the predicted results,and the predicted results are visualized.(4)At last,the overall structure designed in this paper was tested and tested.A small cluster was built in the laboratory.By using the substation battery status monitoring data set,data generation,data access,data processing and data visualization in the overall structure were realized.Through the experimental test of the battery module data of the substation,the CPU and memory utilization rate of the machine are tracked for a long time.Through the analysis of the data obtained from the tracking,it is concluded that the system can maintain reliable and stable operation for a long time,which verifies the feasibility of the overall architecture designed in this paper.