Design And Development Of Microgrid Big Data Platform Based On Cloud Computing Technology

The existing online monitoring and management systems of the microgrid mostly use a relational database to store operational data,and analyze and process data using a single-machine operation mode with limited computing resources.In the face of growing microgrid operational data,existing microgrid management systems cannot meet the storage and computational analysis requirements of large-scale operational data.At the same time,different microgrids are independent and dispersed,and data sharing between multiple microgrids cannot be realized.Based on the parallel computing technology and distributed storage technology in cloud computing technology,this paper combines the front and back of the mainstream Web and network communication technology to design and develop the microgrid big data platform.Firstly,it introduces the research background and significance of this topic,analyzes the status quo of cloud computing research and its application status in microgrid,and expounds the front and back of the Web,parallel computing and distributed required to build microgrid big data platform.The principle of storage,high-performance NIO communication technology.Secondly,the micro-grid big data platform analyzes the functional requirements from four aspects of data acquisition,storage and retrieval,analysis and computation visualization,and designs the technical solutions for different functional requirements.Aiming at the data acquisition problem,a real-time data transmission and collection method based on Netty is proposed and implemented.For the data storage and retrieval problem,the MySQL concurrent storage,index and transaction control design and development,design and implementation of the Solr-based HBase secondary indexing scheme.Aiming at the analysis and calculation requirements,the offline calculation of random forest parallelized photovoltaic power generation prediction and the flow calculation business function of real-time power generation statistics of multi-microgrid are realized.For the visualization needs,the SVG technology is combined with the power system public information model CIM to realize the Web visualization of the electrical circuit,and Vue.js realizes the visualization of the microgrid big data analysis results.Finally,the feasibility of the real-time data collection method is verified by analyzing the throughput,response time and receiver stack operation of the Netty-based microgrid real-time acquisition method.After testing the concurrent performance of multiple database connection pools,BoneCP is selected as the storage connection pool to verify that multiple consumers can perform concurrent read and write tasks to improve storage efficiency.By comparing the performance impact of different column family numbers on the distributed storage system HBase,it is verified that the HBase table uses 1 to 3 column family numbers as the rationality for storing massive microgrid data.After analyzing the random forest parallelization algorithm to predict the average absolute error and acceleration ratio of photovoltaic power generation,it is verified that the random forest parallelization algorithm has good prediction accuracy and acceleration ratio.