Automatic Generation of Real Power Transmission Grid Models From Crowdsourced Data

Abstract

Listen

Translate article

Real models of electrical transmission grids are difficult to obtain. The process of generating such models from unstructured and incomplete data is tedious, and the resulting models are rarely updated. This paper proposes a novel approach for automatically extracting power-relevant data from the public and unstructured crowdsourced OpenStreetMap (OSM) and for generating topology and simulation-ready models of real transmission grids based on the relation between different grid elements such as power lines and substations. Our approach uses spatial analysis and minor assumptions to periodically generate transmission grid models based on the latest OSM data for every country on the planet. A comparison of our generated power grid models with official data from 14 countries reveals accuracy levels between 31% and 94%, caused by the varying availability of OSM data for different countries. Since the crowdsourced data is continuously improving, the automated and periodical model generation approach extends the models with new power circuits as the quantity and the quality of the OSM dataset increases. We provide a platform to access our generated models at opengridmap.org. This paper describes our model generation method, presents our data access platform and evaluates the accuracy of our topological models for selected countries.