电力系统动态无功优化并行算法及其实现.pdfVIP

2009 年 2 月 电 工 技 术 学 报 Vol.24 No. 2 第 24 卷第 2 期 TRAN SACTION S OF CHINA ELECTROTECHNICAL SOCIETY Feb . 2009 电力系统动态无功优化并行算法及其实现 缪楠林 刘明波 赵维兴 (华南理工大学电力学院 广州 5 10640) 摘要 由于全天24 个时段的动态无功优化模型需要考虑变压器 接头和电容器投切开关的 允许动作次数限制，当采用引入离散惩罚机制的非线性原对偶内点法求解时，其修正方程的维 数会随系统规模的增大而急剧增大，但不难发现其修正方程系数矩阵具有对角加边 块结构， 可将其解耦为25 个低维线性方程 。从而提出一种粗粒度的并行计算方法，并在基于消息传 递接口 （MPI）机制的并行计算环境下实现。将所提算法应用于一个实际的 14 节点和IEEE 118 节点系统的实践表明，它能够有效地提升计算速度，在大型电力系统中有着良好的应用潜 力。 关键词：动态无功优化 非线性原对偶内点法 离散惩罚 并行计算 消息传递接口 中图分类号：TM77 Parallel Algorithm of Dynamic Reactive Power Optimization and Its Implementation M iao Nanlin L iu M ingbo Zhao Weix ing (South China University of Technology Guangzhou 5 10640 China) Abstract When a dynamic reactive power optimization problem in next 24 hours is solved by nonlinear primal-dual interior-point algorithm incorporating discretization penalty, the reduced correction equation ’s dimension increases rapidly with enlargement of system size because the limits for switching operations of capacitors and transformer load tap changers should be considered. But it is found easily that the coefficient matrix of the reduced correction equation has block bordered diagonal structure and hence the correction equation can be accurately decomposed into 25 sets of low-dimension linear equations. Therefore, a coarse-grained parallel algorithm is proposed and implemented under message passing interface (MPI) based parallel environment . Numerical results on a real 14-bus and IEEE 118-bus systems demonstrate that the proposed algorithm can accelerate computational process obviously and has application potential in large-scale power syst