Data-driven topology identification in low-voltage distribution networks: A framework-based analysis and future perspectives

The integration of renewable energy sources and the growing complexity of electrical grids have intensified the need for accurate topology identification to ensure system reliability. This work formalizes the topology identification problem through a structured framework, motivated by challenges arising from large-scale blackouts and the expansion of distributed generation. It places a specific focus on low-voltage networks, where sparse observability presents the most significant difficulties, and structures the problem around four critical scopes: bus connectivity, line parameter estimation, substation allocation, and phase identification. Within this framework, existing methods are systematically categorized into Graph Theoretic, Statistical & Correlation, and Machine Learning paradigms, and mapped to their applicability across the four scopes. The analysis highlights a methodological shift from traditional model-based techniques toward data-driven and hybrid approaches. To enable direct comparison across heterogeneous methods, a unified accuracy representation is proposed to transform diverse performance metrics into a comparable percentage-based form. This facilitates a consistent and standardized assessment of how different methodological paradigms address key low-voltage topology identification scopes. This work provides researchers and practitioners with both a structured analytical perspective and a foundation for future developments in topology identification.