A data-driven approach for clustering extra high voltage buses: A case study on the Italian transmission network

The integration of renewable energy sources and the decommissioning of traditional fossil-fuel power plants have introduced new complexities in controlling and monitoring modern transmission networks, particularly regarding voltage regulation. Conventional zoning methods, which are heavily based on system topology, may not adequately address the data-driven demands of these evolving networks. This study proposes a novel, data-driven clustering approach based solely on voltage measurements to define operational zones within a transmission network. To develop and validate this approach, Principal Component Analysis was employed to reduce the dimensionality of high-resolution voltage measurements. Voltage-coherent zones were then formed through Ward's Hierarchical Agglomerative Clustering, applied to the transformed data. The methodology was applied to the Italian Transmission Network, and the clustering results were benchmarked against traditional, topology-based zoning methods used for Secondary Voltage Regulation. The data-driven approach demonstrated strong alignment with topology-based results, validating its effectiveness. This study can provide Transmission System Operators with a flexible, topology-independent tool for defining voltage regulation areas. By leveraging voltage measurements and unsupervised learning techniques, this approach has the potential to enhance grid stability and inform adaptive voltage control strategies, ultimately meeting the evolving demands of modern, renewable-rich transmission networks.