PQ分解法潮流收敛性及收敛速度的新探讨

New exploration on the convergence property and the speed for FDPF algorithm

全面分析比较了PQ分解法潮流计算中不同B′、B″阵参数组合及计算过程对收敛性的影响,分别提出影响收敛性和迭代次数的主要参数。同时考虑打开或建立B′、B″阵数据文件的个数、形成因子表或三角分解的次数、完成前代和回代过程的方式等对PQ分解法收敛速度的影响。通过上述分析可知,在B′阵中去掉参数c,在B″阵中去掉参数r基本可保证对各个系统均有较好的收敛性;去掉B′阵中参数k可减少迭代次数;B′、B″阵的其余参数对收敛性及收敛速度的影响非常小,几乎可忽略不计。因此可知,B′、B″阵在仅保留参数电抗的最简形式下,不但可确保系统的收敛,而且可使潮流计算的速度最快。以IEEE-118节点系统为例,B′、B″阵在最简形式下的计算时间比迭代次数最少的计算时间快约22%。

For the fast decoupled power flow（FDPF）algorithm,the effects of different combinations of paraters in matrices B′and B″and the different calculating methods on the convergence were analyzed and compared in this paper.The main parameters affecting the convergence and the iteration numbers were presented.Furthermore,the numbers of the data files opened or built for storing matrices B′and B″,the numbers of generating the factor tables or triangular factorizations,and the ways of accomplishing the forward and backward substitution were considered,and their effects on the converging speed were investigated.Based on the analysis above,agood convergence can be obtained when cancelling the parameters c in matrix B′and rin matrix B″,and the iteration numbers can be greatly reduced when cancelling the parameters kin matrix B′.However other parameters in matrices B′and B″have almost no effect on the iteration numbers.So,we can know that not only the system＇s convergence can be guaranteed but also the calculating speed would be the quickest under the simplest form for matrices B′and B″.Taking IEEE-118 bus system as an example,the calculating time with the simplest form for matrices B′and B″is about 22%faster than the time with the least iteration numbers for matrices B′and B″.