Identifying 6-Cluster Microgrid Resilience using Enhanced Partitioning Algorithm

Abstract

In light of the growing effects of severe weather on power distribution network (PDN), enhancing resilience is critical. This research presents an advanced clustering method, referred to as the Enhanced Partitioning Algorithm (EPA), designed to establish the limits of microgrids (MGs) within a multi-microgrid (MMG) system. Unlike traditional methods, this approach partitions the PDN into 6 distinct microgrids to improve reliability. The power distribution systems are represented by nodes (buses) and edges (connections), and the analysis includes computation of the adjacency matrix, degree matrix, and Laplacian matrix. The EPA technique is a modification of conventional k-means clustering, utilizing grid-specific features such as terminal points for refined partitioning. Global silhouette coefficients (SC) are measured to evaluate the clustering performance. The method is applied to two IEEE benchmark distribution systems: IEEE 33 & 69 test bus systems. The results demonstrate clear, well-defined clusters with SC values exceeding 0.70, highlighting the importance of terminal points and connectivity patterns in supporting decision-making for grid partitioning. Analysis on the SC of the EPA is compared with the current method. The EPA approach offers researchers and practitioners an effective tool for enhancing the resilience and reliability of modern power grids.