基于伪量测型协调变量的分布式状态估计算法

Distributed State Estimation Based on the Pseudo-measurement Coordination Variables

电力系统状态估计在实际工程中通常由各控制中心采用区域状态估计模式独立完成。考虑到在电网分层分区的控制管理模式下,多样化的电网信息本身就是按照分层分区的分布式模式采集的,分布式状态估计无疑是更加适应该体系的状态估计运行模式。利用集中式优化问题的一阶KKT条件分解技术,实现状态估计任务的分解。为保证在边界上相邻分区状态估计结果的一致性,引入描述边界协调项的等式约束。另外,将上述边界等式约束处理成零功率注入伪量测,以简化分区之间的数据交换,便于重用已有状态估计器。为了尽最大可能维持各个分区计算的独立性,该文给出了协调侧Lagrangian函数梯度的两步式计算方法,并结合Newton-GMRES算法完成求解。最后,通过IEEE标准系统的测试,验证了所提方法的有效性。

Completely independent area state estimation（ASE） mode was adopted by each control center to implement the whole system calculation in practice. However, taking into account the grid under the hierarchical multi-area control and management structure, a variety of grid information itself was collected in accordance with the hierarchical multi-area distributed mode. Distributed state estimation（DSE） was undoubtedly more suitable to the current architecture. The decomposition of state estimation was implemented by the first-order KKT conditions decomposition technique with integrated optimization problem. In order to guarantee the consistency of SE results on the boundaries of adjacent partitions, the equality constraints to describe the boundary coordination entries were introduced. Moreover, in order to simplify the data exchange between the partitions and to facilitate the reuse of the existing state estimator, the boundary equality constraints above were treated as zero-injection pseudo-measurements. In addition, for the purpose of maintaining the calculation independence of each partition as possible as it could, a two-step method was proposed to calculate the gradient of the Lagrangian for the coordination side and to complete the solving with Newton-GMRES algorithm. At last, the effectiveness of the proposed method was verified through the IEEE standard test suites.