Parallel Calculation Applied In Research On The Power Flow And Reliability Of Power System

With the fast development of the power grids, the data computed in analyzing several targets of the power grids are growing time and again. Parallel program can mostly improve the process speed of analyzing the power grids with accuracy. So, there are more and more research on application of parallel program in power flow and reliability analyzing of the power grids.This paper deeply researches the principle,approach of parallel calculation and its hardware and software constitution. We composed Beowulf Cluster with Ethernet,usual PC and software.This paper research on how to improve parallel performance of AC power flow calculation for large-scale power grid by use of preconditioned conjugate gradient (PCG) method is conducted, and a new parallel processing approach, named as the parallel node distribution method, is proposed in which taking the node as unit the data of nodal admittance matrix and nodal output are assigned to each several processors and the remaining parts of calculation and processing are completed in these processors. The analysis and comparison results of various calculation examples show that when PCG method is used in AC power flow calculation of large-scale power grid, the proposed parallel node distribution method is dominant in parallel performance, thus it is available for reference in the application of PCG method in AC power flow calculation of large-scale power grid.Besides, this paper also research the application of Ford-Fulkerson approach in Multi-Area reliability studies, raise a kind of parallel method for the combination process of equal states. That is, equation of each district will be done in every processor, and then divide the equal states of every area into two states with the condition of whether the capacity of the generator satisfies the capacity of the load, and the combination of the state will be handled in every processor, and Ford-Fulkerson approach will be used in each processor to calculate the targets of reliability of each area. Through analyzing the parallel performance used in the combination of every area's states, we know that it can improve the calculated speed fast.