半定规划优潮流的并行计算方法  

A Parallel Computational Method for Optimal Power Flow in Semidefinite Programming

作　　者：田君杨[1] 韦化[1] 白晓清[1] 

机构地区：[1]广西电力系统最优化与节能技术重点实验室(广西大学),广西壮族自治区南宁市530004

出　　处：《电网技术》2014年第1期175-180,共6页Power System Technology

基　　金：国家重点基础研究发展规划项目(973项目)(2013CB228205);国家自然科学基金项目(51167001)~~

摘　　要：优潮流的半定规划松弛方法可提高优解的质量,但存在计算时间长的问题。为此,提出了半定规划优潮流的并行计算方法。该方法对半定规划原始-对偶内点法中的Schur方程采用并行方式进行计算,方面,将系数矩阵各行的计算量分配到不同的处理器上完成,以加速方程的形成;另方面,对系数矩阵进行并行Cholesky分解,从而加速方程的求解。IEEE-118节点和IEEE-300节点2个系统的计算结果表明:所提并行算法的加速效果明显,并保持了内点法的良好收敛性,为半定规划优潮流的工程应用提供了可借鉴的经验。The semidefinite programming （SDP） relaxation of optimal power flow can improve the quality of solutions, but it needs a long time computation. A parallel method to obtain a speed-up for the SDP optimal power flow is proposed. The method takes parallel computation for the Schur equation in SDP primal-dual interior point method. On the one hand, each row of its coefficient matrix is assigned to each processor for calculations, which speeds up establishing the equation. On the other hand, the Cholesky decomposition is implemented in parallel, which speeds up solving the equation. Numerical simulations on IEEE 118 and IEEE 300 systems show that the parallel algorithm has evident accelerating effect and keeps good convergence of the interior point method. The proposed method provides the meaningful experience for engineering applications of the SDP optimal power flow.

关 键 词：最优潮流 半定规划 内点算法 并行计算 

分 类 号：TM71[电气工程—电力系统及自动化]