University of Twente Student Theses


Creating building energy prediction models with convolutional recurrent neural networks

Muller, M. (2019) Creating building energy prediction models with convolutional recurrent neural networks.

[img] PDF
Abstract:Being able to create accurate building energy predictions models can allow for more efficient energy production and save resources. Creating accurate building energy prediction models is a difficult problem because of the many external factors that can influence it, for example the behaviour of people, the weather and electric vehicles. To tackle this problem we will attempt to create building energy prediction models with new techniques in machine learning. We will investigate the multi-input and multi-output inferencing approach, then investigate this approach in combination with convolutional operations. We will investigate the Attention Mechanism to attempt to further enhance the model. The empirical results show promising results. This leads us to new insights into how to build more accurate building prediction models with these techniques.
Item Type:Essay (Bachelor)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Computer Science BSc (56964)
Keywords:Recurrent neural networks, Gated recurrent unit, Convolutional neural networks, Attention mechanism, Building energy prediction
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page