Faculty of Computer Science

Research Group Theoretical Computer Science

Oberseminar: Heterogene formale Methoden

Abstract:

Energy system optimization models play a crucial role in enhancing system performance, reducing environmental impact, and facilitating economic expansion planning of energy systems. However, the sheer complexity of heterogeneous and multi-dimensional input data presents computational challenges that hinder efficient execution. Existing unsupervised methods for temporal reduction in energy system optimization models (ESOM) have been used to address this issue but often result in significant accuracy loss due to the exclusion of ESOM output. To overcome this limitation, this study proposes a novel supervised deep learning-based framework. This framework aims to reduce the temporal sub-dimension (timesteps) of ESOM input data while minimizing the trade-off with the accuracy of the ESOM model. By leveraging the power of deep learning techniques, the proposed framework holds the potential to optimize energy system models efficiently and effectively.