Inference of Distribution Grids Based on Crowdsourced Grid Data and Drone Imagery

Abstract

Distribution System Operators (DSOs) face several challenges in managing comprehensive and up-to-date models of distribution grids. To address these problems, we propose a crowdsourcing framework for collecting grid devices. We also provide an inference approach for generating topological models of the distribution grids. Since distribution cables are often underground, we use spatial data analytics on the collected data in combination with other open data sources to infer the topology of the distribution grid. Additionally, to increase the quality of crowdsourced data, we propose a cost-effective approach for collecting and detecting grid elements in urban areas using commercial drones with an RGB camera. To evaluate our approach, we organized a crowdsourcing campaign to map and infer a district in Munich, Germany. The results are compared with the ground truth of the distribution system operator. Our results report a precision of up to 82 percent and a recall of up to 65 percent for the correctly crowdsourced grid devices. We also observe that the inferred models achieve a power length accuracy of 88 percent compared to the ground truth. We evaluated the detection of solar panels from aerial imagery by conducting field experiments, showing precision and recall levels of 68 and 69 percent, respectively.