Edge Computing Architecture for Optimal Settings of Inverse Time Overcurrent Relays in Mesh Microgrids

This paper presents a novel edge-computing-based architecture for optimal inverse time overcurrent relays installed to protect mesh microgrids (MGs) with distributed generation. The procedure employs graph theory to automate the detection of network changes, fault locations, and relay pairs in an MG. In addition, an automated process obtains the initial protection settings based on the operating conditions of the MG. Furthermore, the Continuous Genetic Algorithm (CGA), Salp Swarm Algorithm (SSA), and Particle Swarm Optimization (PSO) were implemented to determine the optimal protection settings to obtain better coordination between primary and backup protection relays. These processes were implemented using PowerFactory 2024 Service Pack 5A and Python 3.13.1. The proposal was validated in 68 operating scenarios that considered the islanded and connected operation modes of the MG, charging and discharging cycles of electric vehicle stations, and the presence or absence of photovoltaic generation. The overcurrent protection relays were organized into 100 primary–backup relay pairs to ensure proper coordination and selectivity. The total miscoordination time (TMT) index was used to measure when all pairs of relays were coordinated, with a minimum time close to zero. The results of the graph theory show that all the meshes, fault locations, and relay pairs were identified in the MG. The approach successfully coordinated 100 relay pairs across 68 scenarios, demonstrating its scalability in complex real-world MGs. The automation process obtained an average TMT of 12.2%, while the optimization obtained a TMS of 91.6% with the CGA, and a TMT of 99% was obtained with the SSA and PSO, demonstrating the effectiveness of the optimization process in ensuring selectivity and appropriate fault clearing times.