基于分区解耦的配电网状态估计的分布式并行计算

以全量测配置为边界对配电网馈线进行解耦分区，把分区作为独立的分析单元对状态估计进行分块处理，并采用基于支路电流的量测变换技术对配电网的单分区进行状态估计。同时，构建了分布式并行计算的分层结构，引入基于消息中间件ZeroMQ技术，采用不同类型套接字的组合实现分布式系统内部的N-N高效通信。并在分布式并行计算平台上以分区状态估计为子任务，实现全网状态估计的分布式并行计算。算例分析表明：在配电网规模达到500节点及以上时，采用所提出的状态估计分布式架构进行计算具有明显的速度优势。

Parallel and distributed computing for an area-decoupled state estimation method for distribution systems

Considering the scale of current distribution systems, parallel and distributed computing is addressed. Which parallel technology and architecture to take and how to divide the computing unit are two key issues of distributed computing. A hierarchy of distributed platform is built based on ZeroMQ, which can provide N-N connection by different kinds of socket. To support the parallel computing, a feeder is decomposed into subareas which are area-decoupled, so that we can regard each subarea as an independent estimation unit. At the same time, the branch-current-based method is proposed to estimate the state of each subarea. As the state estimation (SE) of a subarea being a sub-task, we can finally achieve the SE of the whole distribution system. The calculation example shows that with the proposed platform, SE for a 500-node system takes a significant speed advantage than the traditional ones.