The content discusses the importance of thermal power flow (TPF) in 4th generation district heating grids and proposes a novel, efficient scheme to generate training data sets. By using a proxy distribution approach, the algorithm omits iterations needed for solving heat grid equations, resulting in faster training set generation times. The study shows that this new approach can reduce computation times by up to two orders of magnitude compared to traditional methods without sacrificing accuracy. Additionally, learning TPF with a training data set outperforms sample-free physics-aware training approaches significantly.
The article highlights the transition from 3rd to 4th generation district heating grids, emphasizing low supply temperatures and the integration of low-carbon heat sources. It discusses classic TPF computations using iterative algorithms like DC and NR, contrasting them with faster learning-based approaches using neural networks. The proposed algorithm aims to streamline the model-building process for TPF calculations by efficiently generating relevant training data sets.
Para Outro Idioma
do conteúdo original
arxiv.org
Principais Insights Extraídos De
by Andreas Bott... às arxiv.org 03-19-2024
https://arxiv.org/pdf/2403.11877.pdfPerguntas Mais Profundas