Determination of sectionalising strategies for parallel power system restoration: A spectral clustering-based methodology

Parallel power system restoration (PPSR) accelerates the restoration of a system in complete blackout by restoring isolated sections (islands) of the network in parallel. These islands are defined during the preparation stage of PPSR using a sectionalising strategy that considers system information collected after the blackout and must satisfy multiple PPSR constraints. This paper introduces a methodology based on spectral clustering that, in contrast to existing approaches, uses the physical and inherent properties of the network to determine a suitable solution. An undirected edge-weighted graph is initially constructed based on the electrical distance between buses, and constraints related to transmission line availability and cranking groups are included by modifying the edge-weights of the graph and using a subspace projection. This graph is then used to define islands that have strong internal connections but weak external connections, whilst satisfying the following constraints: blackstart availability, load-generation balance, voltage stability and the ability to monitor synchronisation between adjacent islands. Simulation results for the Institute of Electrical and Electronic Engineers (IEEE) 39- and IEEE 118-bus test systems demonstrate the ability of the methodology to define a strategy that creates strongly connected islands. Additionally, they indicate that the new approach determines solutions that have larger ratios between the inter-cluster electrical distance and the intra-cluster electrical distance for larger systems.