This paper proposes a multi-agent, zone-based protection scheme for modern radial smart distribution grids with high penetration of Distributed Energy Resources (DERs). Conventional protection systems, designed for unidirectional fault currents and static topologies, struggle in dynamic, inverter-dominated networks. The proposed solution manages faults using local agents assigned to each zone. A universal three-phase Fault Detection Module (FDM), based on a compact set of features extracted from the mean and superimposed components of three-phase voltage and current signals, is presented to reliably distinguish faults from pseudo-faults under varying conditions. Fault section identification is achieved through a lightweight, communication-efficient scheme using binary fault flags exchanged within each zone. Furthermore, a backup mechanism is introduced to ensure resilience against relay failures, and a Severe Fault Detection Module (SFDM) is designed to isolate critical faults immediately, bypassing coordination logic. The scheme is validated using EMT simulation tool of DIgSILENT on the IEEE 33-bus system, along with MATLAB-trained ANNs, across more than 100,000 fault cases and 5000 pseudo-fault scenarios. Results demonstrate over 95% classification accuracy, a 21.5 ms decision time, and robust performance in both islanded and grid-connected modes, confirming its effectiveness and scalability for real-world applications.
Decentralised Zone-Based Protection of Radial Smart Grids Using a Multi-Agent Scheme
Siano P.;
2026
Abstract
This paper proposes a multi-agent, zone-based protection scheme for modern radial smart distribution grids with high penetration of Distributed Energy Resources (DERs). Conventional protection systems, designed for unidirectional fault currents and static topologies, struggle in dynamic, inverter-dominated networks. The proposed solution manages faults using local agents assigned to each zone. A universal three-phase Fault Detection Module (FDM), based on a compact set of features extracted from the mean and superimposed components of three-phase voltage and current signals, is presented to reliably distinguish faults from pseudo-faults under varying conditions. Fault section identification is achieved through a lightweight, communication-efficient scheme using binary fault flags exchanged within each zone. Furthermore, a backup mechanism is introduced to ensure resilience against relay failures, and a Severe Fault Detection Module (SFDM) is designed to isolate critical faults immediately, bypassing coordination logic. The scheme is validated using EMT simulation tool of DIgSILENT on the IEEE 33-bus system, along with MATLAB-trained ANNs, across more than 100,000 fault cases and 5000 pseudo-fault scenarios. Results demonstrate over 95% classification accuracy, a 21.5 ms decision time, and robust performance in both islanded and grid-connected modes, confirming its effectiveness and scalability for real-world applications.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


