为解决应用内嵌离散惩罚的非线性原对偶内点法求解离散整数动态无功优化模型时产生的“维数灾”问题,对修正方程用块矩阵解耦的算法做进一步探讨,提出了两次求解修正方程系数矩阵并三角分解从而降低动态无功优化应用于大电网时的数据存储量的新思路,即以时间换取空间(定义为时空转换)。在两个实际系统(14节点和538节点系统)和IEEE118节点系统上的优化计算表明,所提算法既能计算大电网的动态无功优化,又具有较快的计算速度。
In order to resolve the so-called problem of "curse of dimensionality", this paper analyzed dynamic reactive-power optimization by using nonlinear primal-dual interior-point algorithm which introduced discretization penalty and showed the further exploration of block-matrix decoupling method used in the eoefficient matrix of the reduced linear correction equation. Moreover, a realization of exchanging time for space (time-space transformation) by twice computing coeffieient matrix of the reduced linear correction equation and its triangu- lar factorization which can reduce the data storage is given. The numerical results on IEEE 118-bus system and two real power supply systems with 14 nodes and 538 nodes show that the proposed algorithm is suitable for using in large-scale power systems and has fast calculation speed.