Power System Controlled Islanding Using Directed Motif-Based Spectral Clustering: A Novel Complex Network Perspective

Abstract

Intentional Controlled Islanding (ICI) is a wide-area self-healing strategy to prevent power system blackouts. Recent studies integrate ICI within a complex network framework using community detection techniques, thus addressing the complex nature of power systems. The spectral clustering algorithm (SCA) has shown effectiveness in community detection within complex power networks. However, the focus of SCA on undirected networks fails to satisfy the generator coherency constraint. Additionally, it inadequately represents the electrical characteristics required for optimal islanding. This paper implements the ICI scheme via directed community detection, enabling comprehensive community discovery. The process begins with power flow tracing, creating a directed weighted network between generators and loads. To analyse this network, we apply the motif-based spectral clustering algorithm (MSCA) that accounts for both the direction and weight of the edges. Specifically, we define electrical motifs in power networks as high-order subnetworks considering directed weighted connectivity patterns between generators and loads. Numerical simulations on various test cases compare the performance of MSCA and SCA to evaluate the proposed method. According to the results, the SCA based on motifs, as employed in this paper, outperforms traditional SCA using undirected edges as low-order structures. This novel approach increases load restoration and reduces restoration time.