Synthetic Power Network Topology Generation with Geographical Information

Enhancing the resilience and expansion planning of power grids demands test cases enriched with geographical information.However, due to privacy issues and limited data availability, accessing opensource datasets that contain such information remains a significant challenge.This paper presents a synthetic power network topology generation method that systematically incorporates geographical data into existing power grid topologies.By selecting target regions based on distinct grid characteristics, our approach diversifies grid representations and includes a rigorous evaluation process using multi-modal large language models (MLLMs) to ensure output consistency.This approach uniquely maps grid topologies to real-world locations, creating networks that reflect the spatial characteristics of actual power grids.To enhance the applicability of the generated data, we provide an interactive website that allows users to customize connections and substation locations.The open-sourced codes are accessible at https://github.com/SYLLABLE604/PGGKE.