Research On Power System Parallel Fault Calculation Based On Multi-core Cluster

Protection relay setting and coordination is determined by multiple fault calculation and setting calculation in the software which simulates various modes of power system operation. As the complexity of power grid increased, the modes of power system operation are also increased which makes the amount of fault calculation rise, and it hinders the development of Protection relay setting and coordination.The rapid development of computer and Internet promotes the application of parallel technology. Parallel technology is used to improve the efficiency of the fault calculation in this paper. The main contents are as follows:The program of fault calculation is designed by OpenMP(Open Multi-Processing) technology, and it is also simulated in multi-core parallel system. Considering of the shared memory in multi-core computer, the initial information of power grid is introduced into the memory. The parallel threads finish multiple fault calculation simultaneously with the information in the shared memory. Some variables are privatized by the parallel threads to weaken the bad effects of the data competition and ensure the accuracy of parallel fault calculation. The mission of multiple fault calculation is well allocated by the way of dynamic scheduling.The program of parallel fault calculation based on MPI(Message Passing Interface) is designed, and the parallel speedup is acceptable. The structure of the memory is distributed in cluster of PCs. The master of the cluster finishes the serial computation and broadcasts the mission to slaves in the cluster. The fault calculation is handled simultaneously by all PCs in the cluster. The cluster is set up well by the software MPICH which is built in all PCs, and it is also the interface of parallel calculation.The fault calculation of power grid is handled in multi-core cluster by MPI_OpenMP hybrid technology, and the speed of fault calculation is improved significantly. The single hybrid model is adopted when the program is designed. The responsibility of MPI level is dealing with the initial data of power grid, broadcasting the mission of fault calculation and collecting the results of fault calculation. The responsibility of OpenMP level is finishing the process of multiple fault calculation including analyzing the data, modifying the admittance matrix and solving equations. The parallel fault calculation based on MPI_OpenMP is tested by IEEE30 system and an actual power grid, and the parallel speedup is remarkable.The efficiency of fault calculation is improved significantly. The scalability of the parallel fault calculation based on MPI_OpenMP is good, and the parallel speedup can be improved continuously when the scale of the multi-core cluster expands.The above kinds of parallel fault calculation can be applied to different parallel systems. It is necessary for parallel fault calculation to find the suitable technology relaying on the actual environment.