Development of hierarchical two-stage constrained spectral clustering algorithm to enhance power system distribution network resiliency under zonal attacks

The power grid is intended to deliver electric power from large, remote power generation units to end-consumers in residential, commercial, and industrial sectors. The traditional grid design is drastically altering as a result of recent trends such as increased allocation of renewables, distributed energy resources (DERs) with uncertainty in power, and environmental changes that are pushing weather scenarios toward extremes. Reliability and resiliency are quickly evolving as key issues in the power system in order to react quickly and securely to those significant changes. Extreme weather conditions may result in numerous contingency outages, cascade failures, zonal attacks, and widespread blackouts. This paper proposes a framework for effective quantification analysis of zonal attack and formation of two-stage constrained spectral clustering. This research focuses on a two-stage constrained spectral clustering method for managing critical infrastructure in an integrated DER power distribution network. The primary objective is to provide a virtual clustering framework based on the real power flowing of lines in the power distribution network that can deliver a reliable and consistent cluster. The proposed control algorithm is tested on IEEE 123 bus distribution system. Based on the position of sectionalizing switches, the test system is divided into seven zones. Zone-7 and zone-5 attacks are visualized in the MATLAB/SIMULINK (2023a) environment and are found to perform better than the alternatives and provide an effective power balance. The proposed model is also developed in OPAL-RT OP-4512 HIL real-time simulation platform; the zone-4 attack is created, and the clustering results are verified using a real real-time digital simulator. In addition to offline simulations, the performance of the proposed methodology has been verified in real time by running hardware-in-the-loop simulations on the OPAL-RT real-time simulator. Finally, the clustering quality is evaluated using metrics like Ncut, discontinuities, maximum expansion, and ratio of cluster sizes.