The Research Of Architecture Of Distributed Data Mining Based On Grid

There are some research fields need data analysis and process in electrical engineering area, for example, power load forecasting, electrical equipment monitoring on line, real time pricing system in power market, current analysis system and so on. And for those problems need to analyze a large number of data, data mining is one of the current main solutions. The improvement of efficiency and accuracy of data mining can strengthen the data analysis and processing, which may be helpful for solving some problems in electrical engineering.In fact, most of large database is distributed and so data mining is distributed, which leads to the research on distributed data mining is more practical. Currently there are many achievements on the distributed data mining. However, there are also some limitations which are mostly caused by the limitations of current popular distributed computing technology. As a result, the paper do research on the architecture of the distributed data mining, make use of the latest development of distributed computing technology—grid computing, to build an architecture of distributed data mining based on grid.Firstly, in this paper, the current popular grid architecture are discussed and analyzed, as well as the advantages and disadvantages of five sandglass and open grid service architecture, which supports the choice of Web service resources framework as the foundation of distributed data mining architecture.Secondly, according to the feature of distributed data mining, the architecture based on web service resources framework is designed, as well as the workflow under it. In addition, the components of the whole architecture, including data web service resource, computing web service resource, algorithm web service resource, data mining service, are designed respectively.Finally, the various kinds of web service resource are implemented, which means describing the resource properties documents of these web service resources, implementing the manipulate interfaces of these documents, describing the static topology of these web service resources by UML. Furthermore, a virtual local grid is built by installing the grid middle ware on the some virtual machines in a virtual local network generating by the software—Virtual Machine Ware, and on which an association rules mining instance running successfully proves the feasibility of the proposed architecture in...