PROGRESS - Probabilistic graphical energy systems - Innovative methods for the holistic consideration of uncertainties in planning problems of the energy system

Abstract

With the novel cooperation of two partners from different disciplines, the overall goal is the development and implementation of fundamentally new analytical capabilities for future energy system considerations. The combination of Princeton University's expertise in solving large scale optimization problems with Fraunhofer IEE's experience in energy system analysis will allow the development of a new class of modeling and optimization frameworks that can make decisive contributions in industry and the public sector. The methodological approach, which has been proven in computer vision applications, is based on graphical probabilistic models and differentiable solvers using proxy functions. While in computer science, the project can make important contributions to optimisation research, applied energy research can obtain innovative tools for solving problems that have not been solved so far. In addition to the open-source exploitation of the scalable optimization approaches, the endogenous quantification and visualization of uncertainty relationships is a major contribution. Associated partners from science and industry should ensure the relevance and added value from the user's point of view with their requirements and questions.