Accelerating The Formation Of State Space Equations Of Power Systems Based On BICGSTAB And ILUTP Technologies

With the emergence of large-scale interconnected power systems,the issue of small interference stability of the system has attracted much attention.Among them,the eigenvalue analysis method,which focuses on solving state space equations,has gradually fallen into a "dimensional disaster" with the continuous expansion of the matrix scale in power systems.In order to reduce the calculation time of small-interference stability analysis of large-scale power systems,this paper studies the rapid formation method of state space matrix:For the plug-in modeling technology,analyze the formation of the state matrix;The incomplete LU factorization(ILUTP)method with double thresholds in the preprocessing technique is used to adjust the position of non-zero elements in the correlation matrix to convert the matrix into a diagonally dominant form;Then,the double conjugate gradient stabilization method(BICGSTAB)in the Krylov subspace is used to avoid the operation of matrix transposition and iteratively solve the processed large sparse matrix.In this paper,ILUTP and BICGSTAB algorithms are used in combination to speed up the formation of the state matrix.Secondly,using the algorithm characteristics of ILUTP and BICGSTAB,parallel computing based on Open MP technology is realized;Parallel computing can divide tasks into multiple non-interfering subtasks and assign them to multiple threads for common execution,thereby reducing computing time.This article uses the parallel computing function of Open MP technology for some DO loop statements and independent program segments in the algorithm to improve the overall computing efficiency.In addition,in order to take full advantage of the sparseness of the correlation matrix,the storage of the matrix uses a combination of triplet technology and row compression sparse storage technology(CSR);triplet technology is suitable for the establishment and export storage of matrices,and row compression storage is suitable Operations between matrix vectors.The flexible use of the two in this paperreduces the memory footprint during the formation of the state space matrix.Finally,this article uses Fortran language to write the program in the environment of Visual Studio 2010 platform and Intel Fortran 12,and execute it on a computer with an eight-core processor.Based on two actual examples,the method in this article is based on PMT and PMT respectively.The original method and other improved methods based on PMT were compared and analyzed,which verified the feasibility and effectiveness of the method in this paper.The results show that the combined use of ILUTP,BICGSTAB,CSR,and Open MP can speed up the formation of the state space of large power systems.