I am deeply passionate about the future of buildings and how computational technologies will be integrated into the design process and controls of buildings. My main focus is on developing solutions for improving resource efficiency in the built environment. I draw expertise from core building design concepts, ICT, controls, simulation, and data analysis with the goal of delivering novel solutions to complex technical challenges.

“Control-oriented deep neural networks for building dynamics prediction”

Currently, I am a PhD candidate at the Department of Energy and Process Engineering since January 2020. The main goal of this project is to develop generalized control-oriented deep neural networks which can be used as estimators in optimal control for buildings. At its core, the research delves into the efficacy of deep neural networks rooted in the encoder-decoder architecture, specifically for predicting multi-zone, multi-output, multi-step ahead building dynamics responses based on given control inputs. Another dimension explores the transfer learning advantages, focusing on how pre-trained networks can be adeptly fine-tuned for predicting building dynamics across varied climates.

On the practical objectives side, I aim to develop a comprehensive tool or framework that supports the streamlined creation and optimization of modular sequence-to-sequence control-oriented deep neural networks for dynamic system predictions. Another objective is to develop an effective methodology or tool that generates synthetic datasets, ensuring that control-oriented neural networks for building dynamics prediction are trained and validated in an environment mirroring real-world scenarios.