Network-partition-based source-load-collaboration voltage regulation with optimal communication routing in active distribution network

To enhance voltage quality in active distribution networks (ADNs), a large number of distributed energy resources and loads are integrated for voltage regulation. However, their large scale deployment poses significant challenges to communication and control systems. To address this issue, a network-partition-based source-load collaborative voltage regulation method with optimized communication routing is proposed. First, a graph convolutional network (GCN) integrated with a long short-term memory (LSTM) model is developed to rapidly evaluate the voltage regulation capability (VRC) of each node, enabling the partition of the network into independent Voltage Support Regions (VSRs) and Voltage Joint Support Regions (VJSRs). For VSRs, a distributed voltage average estimation method is introduced to classify nodes into two regulation states, followed by a two-stage distributed sensitivity-based droop control strategy for fast voltage regulation using only local communication links. For VJSRs, a centralized sensitivity-based multi-VSR coordination control is established to realize collaborative voltage regulation. In addition, an optimized communication routing algorithm is designed for centralized control, considering cyber-physical coupling factors such as communication delay and topological uncertainty. Simulation results demonstrate the effectiveness of the proposed method.