基于轻量级实现的约束分支搜索拓扑识别流计算方法

Accurate topological awareness is critical to the stability of low-voltage distribution networks (LVDNs). However, traditional impedance-based topology restoration assumes accuracy that is often unattainable due to impedance data inaccuracy. Given LVDN sensor quality, robustness against data quality issue is crucial. Additionally, the integration of distributed energy resources (DERs) is expanding. Identifying their locations is necessary for effective load management and decreasing utility loss. Conventional identification methods rely on centralized data processing. However, they are limited due to increased storage and computational demands. This paper presents a novel approach employing constrained branching search within a stream computing framework, tailored for radial LVDNs. The proposed method uses node connection (NC) restrictions to recover topology. These constraints are based on the radial LVDN physical model. Additionally, a mathematical model for plug-in PV locations is introduced. We design CommuniDispatch, a lightweight implementation stream computing framework integrating our topology identification method. Enhanced by a Latin hypercube sampling-based recursive bound & search (LHS-RBS) algorithm, it significantly amplifies computational efficiency. Our experiments on diverse radial LVDNs validate the method's accuracy in topology identification and robustness against data quality issues, along with plug-in PV location and the computational efficiency of the LHS-RBS.