After a blackout, it is essential to restore the blackout area rapidly to minimize possible losses. In parallel restoration, the blackout area is first partitioned into several subsystems, which will then be restored in parallel to accelerate the restoration process. In order to ensure restoration reliability, each subsystem should have enough generation power and satisfy a set of constraints before triggering the parallel restoration process. This paper models this as a constrained optimization problem and proposes a partitioning strategy to solve it in three steps. In the first step, some existing methods and expert knowledge are used for initialization of the partitioning process. The second step ensures the satisfaction of modeled constraints. The third step operates greedily to find suitable partitions for parallel restoration. The proposed strategy is implemented and evaluated on IEEE 39- and 118-bus power systems. Evaluation results show that it provides adequate subsystems for parallel restoration. Unlike some existing partitioning strategies, the proposed strategy can be used to partition a power system into multiple subsystems in a single execution.

中文翻译：

---

基于聚集聚类的网络分区用于并行电力系统恢复

停电后，必须快速恢复停电区域，以最大程度地减少可能的损失。在并行还原中，停电区域首先被划分为几个子系统，然后将其并行还原以加快还原过程。为了确保恢复可靠性，每个子系统在触发并行恢复过程之前都应具有足够的发电能力并满足一组约束。本文将此模型建模为约束优化问题，并提出了一种分区策略来分三步解决。第一步，使用一些现有方法和专家知识来初始化分区过程。第二步确保模型约束的满足。第三步贪婪地操作以找到合适的分区以进行并行恢复。所提出的策略是在IEEE 39总线和118总线电源系统上实施和评估的。评估结果表明，它为并行恢复提供了足够的子系统。与某些现有的分区策略不同，所提出的策略可用于在一次执行中将电力系统分区为多个子系统。