Font Size: a A A

Research On Hadoop Distributed Computing Platform For Power Application

Posted on:2017-05-17Degree:MasterType:Thesis
Country:ChinaCandidate:Y WangFull Text:PDF
GTID:2309330488983695Subject:Computer technology
Abstract/Summary:PDF Full Text Request
With the acceleration construction of the smart grid, the grid has expanded rapidly, the structure of power system has be more complex, and the data has increased significantly. The emergence of this series of changes makes the development of computation software used in power system encounter a bottleneck. Hence, building a safe, stable and efficient management software has become a top priority.Since the concept of cloud computing is proposed, it has been studied and applied into several areas by major Internet companies and laboratories. Hadoop is an open source platform which belongs to cloud computing processing tool with efficient parallel computing power and stability of distributed storage capability. In this paper, we study this platform and try to use it to slove various types of complex tasks in distributed power system.Considering the characteristics of computation tasks in distributed power system, the default load balancing strategy in Hadoop can not perform large-scale computing. This paper proposes an improved multi-objective intelligent optimization algorithm-Bacterial foraging algorithm(IMOBFA) to solve load balancing problems with taking into account all kinds of factors of large clusters in actual situation. By using cloud computing simulation software-CloudSim to simulate Hadoop operating environment, we compare the effect of executing massive computation of IMOBFA and the conventional BFA. The experimental results show that the effect of IMOBFA has more obvious advantages.Finally, the small-scale Hadoop cluster is set up in laboratory to do a real test with data of distributed power system from a province of East China. The Hadoop platform with improved load balancing startegy and a distributed platform which has been operating in power system production environment have been compared. When the grid data reaches large enough, the advantage of the Hadoop platform in caculation is outstanding. In addition, since Hadoop platform has the ability about distributed storage, it increase the safety of data storage, which provides basic of big data analysis for building future smart grid.
Keywords/Search Tags:Smart Distribution Grid, Hadoop Platform, Multi-Objective Bacteria Foraging Algorithm, Load Balancing
PDF Full Text Request
Related items