Optimal Network Clustering in Edge Computing Architecture for Distribution Network Monitoring

The management of large measurement sets within power systems for state estimation (SE) imposes a communication burden, consequently leading to delays in SE. To address the above challenges, this paper proposes an optimal edge-computing architecture to provide robust support for SE. Potential edge estimation points (EEPs) are identified, which function as edge servers (ESs), responsible for estimating nodes within clusters. An initial network clustering scheme is established using a novel network evolution concept and electrical distance analysis. To achieve the optimal allocation of EEPs and clustering in alignment with the defined objectives, a modified mutation operator is introduced within a genetic algorithm (GA), tailored to the network's topology. The objectives are designed to streamline the SE process and generate appropriate clusters. A comprehensive analysis of the IEEE 123-bus system has been conducted, revealing a multitude of viable EEP-laden clusters suitable for SE, thereby showcasing the effectiveness of the optimization approach. The solutions outperform the benchmark solutions. Furthermore, the algorithm showcases fast convergence, underscoring its efficiency in addressing the evolving demands of distribution network management.