Research Project OASES

Energy system modelling is the basis for the development and integration of renewable energy (RE) at local, national, regional, continental, and global levels. The resulting energy scenarios are crucial for understanding the contexts in which technologies and energy solutions need to be developed and contribute to being able to optimise future energy systems.

In the field of energy system modelling, there is a lack of holistic approaches that link all steps from the generation of RE input data, potential analysis, RE distribution, time series generation to system models. The required input data are the interfaces to the energy system model in terms of the spatial distribution of renewable energies, considering existing plants, potential areas and resource assessment, as well as the time series generation based on these.

The overall objective of the project is to develop and demonstrate a sustainable AU-EU ecosystem for energy system modelling based on open-source software and freely available data.

Fraunhofer IEE is responsible for the overall coordination of the project and is also researching the distribution of renewable energies under various boundary conditions.

Machine learning methods, especially from the field of deep learning, can be used to automatically extract information from images such as aerial and satellite images. To do this, models are specifically trained for these tasks. For the detection of PV systems in various drone, aerial, and satellite images, a comprehensive model was trained specifically for this project. The description of the method and the final model have been published for free use.

How can a user without programming knowledge use the models?

For this purpose, the open-source QGIS plugin "Deepness: Deep Neural Remote Sensing," developed by the PUT Vision Lab (Computer Vision at Poznan University of Technology), can be used. It offers an intuitive interface for applying image recognition, object regression, and semantic segmentation in the domain of remote sensing. The plugin has been enhanced with a model developed in the project for the detection of PV systems. A deep learning-based model for the segmentation of PV systems in various resolution aerial and satellite images, jointly developed by the Department of Energy Management and Operation of Electrical Networks, the Fraunhofer IEE, and the Council for Scientific and Industrial Research (CSIR) from South Africa, has been integrated so that users can apply it directly to any location on the globe without further programming knowledge.