Virtualized Intelligent Relaying of Smart Grid Over 5G Network

Abstract

Integrating artificial intelligence (AI) and virtualization technology into power systems applications can facilitate the transition from conventional grids to smarter grids. Virtualization of grid application, including AI-based applications, near the edge of the grid reduces communication latency and promotes centralized or decentralized decision-making. The infrastructure of edge or cloud datacenters serves as a robust platform for deploying intelligent virtual applications within smart grids. In this paper, we propose a concept of intelligent virtualized relaying using 5G communication. The concept is demonstrated through a hardware-in-loop setup consisting of an MLP algorithm deployed on the edge server, a grid model in OPAl-RT, and UDP protocol communication of voltage and current information over the 5G network. In 5G communication, latency poses a challenge when a fault occurs in the power system and immediate decision-making is required. In this paper, we attempt to predict faults, which are followed by precursor events or pre-fault signatures, to overcome the latency barrier of 5G communication. Our MLP algorithm achieved an accuracy of 99.9%, enabling it to predict faults based on pre-fault signatures, isolate the fault, and island the load with backup distributed generation (DG). Based on the results obtained from Test-1 and Test-2, the fault was predicted within 90ms and 200ms.