基于CSR直接列写节点导纳矩阵的大电网直流潮流快速计算方法

Fast and Large-Scale DC Power Flow Calculation Algorithm Based on CSR Storage Format via Direct Node-Admittance Matrix Formulation

针对传统的节点导纳矩阵列写并转换为压缩存储时所需计算内存过大、潮流方程求解耗时长的问题,提出了一种按CSR(Compressed sparse row)压缩存储格式直接列写节点导纳矩阵的大电网直流潮流快速计算方法.算法计及多回线的等值,按节点编号顺序计算节点导纳矩阵的上三角非零元素,拼接形成CSR格式的数值、列号、行偏移列向量.从收敛交流潮流中获取支路损耗或估计不收敛网络的支路损耗,将支路损耗作为等值负荷计入对应节点的有功负荷.采用支持CSR格式的PARDISO求解器实现基于稀疏技术的直接法快速求解.IEEE300节点算例与某11827节点实际电网的计算结果表明:所提方法求解速度快、无需迭代、占用内存小,与交流潮流计算结果吻合度高,能够满足大电网快速潮流计算的需求.

In order to solve the problems of large computational memory and time-consuming process of solving the power flow equation when traditional node admittance matrix is listed and converted to compressed storage format,a fast calculation method for DC power flow of large power grids via directly listing the compressed storage format of node admittance matrix is proposed. The algorithm takes the equivalence of multiple branches into account, calculates the upper triangular non-zero elements of the node admittance matrix according to the node number sequence, and creates the value, column number, row offset column vector of the CSR format. The branch loss from the convergent AC power flow or estimate the branch loss via the non-convergent network is obtained, and then considered as the additional equivalent active load on the corresponding nodes. The PARDISO solver supporting the CSR format based on sparse technology is used as the solver. The calculation results of the IEEE 300-node case and the actual power grid of a 11827-node case show the proposed method is fast, non-iteration, low memory consumption, and is highly consistent with the AC power flow calculation results, which meets the needs of fast power flow calculation in large power grids.