Research On Parallel Computing Of Electric Power System Based On Grid

At present, the operation and control of electric power system have turned into an application engineering which relate to a large amount data and imformation. Hence, it is necessary to introduce the grid computing to the application of electric power engineering, in order to explore new method which will make control and decision-making on large network more efficient.In this paper we proposes a grid computing architecture for the electric power system, and a selective algorithm based on multiple QoS constraints for grid task scheduling is presented. A blocked parallel algorithm for the power flow Jacobi matrix equation is designed based on the parallel triangular decomposition. A redundant computing program is introduced; At last, these algorithms are implemented as grid services and deployed at Globus grid nodes. The function verification test is performed in the laboratory. The results show that the electric power flow parallel computing system based on Globus grid platform has the potential for distributed parallel computing, and it is prospective.