Page 39 - InnoRenew CoE International Conference 2021, Healthy and Sustainable Renovation with Renewable Materials
P. 39
N Supporting EDS Building Optimisation
Tamás Storcz 1, Kristóf Roland Horváth 2, István Kistelegdi 3, Zsolt Ercsey 4
1. Department of Systems and Software Technologies, University of Pécs, Faculty of Engineering and Information Technology, 7624
Pécs, Boszorkány street 2, storcz.tamas@mik.pte.hu
2. Marcel Breuer Doctoral School, University of Pécs, Faculty of Engineering and Information Technology, 7624 Pécs, Boszorkány
street 2, horvath.kristof.roland@mik.pte.hu
3. Energy Design Research Group, University of Pécs, Faculty of Engineering and Information Technology, 7624 Pécs, Boszorkány
street 2, kistelegdisoma@mik.pte.hu
4. Department of Systems and Software Technologies, University of Pécs, Faculty of Engineering and Information Technology, 7624
Pécs, Boszorkány street 2, ercsey.zsolt@pte.hu
Energia Design Synthesis (EDS) is a method to design buildings considering both energy efficiency while
offering the best comfort. Earlier as part of the method, based on geometry, energy and climate-related
rules, all potentially optimal building configurations were generated, and complex energy and comfort
simulations were performed to identify the optimal design solution for some simple classes of problems.
For the investigated simple classes, serious limiting architectural rules had to be considered to evaluate
each potentially feasible situation.
To broaden the range of the covered cases, the EDS method is extended, and some artificial intelligence-
based techniques are applied. Here, based on the dataset of some previous complex dynamic thermal
simulations performed, an artificial neural network is built and optimised. On the elaborated dataset,
precision and recall are determined. With the help of supervised learning, further cases with not yet
investigated input parameters can be predicted and, in addition, future optimization can be carried out
with considerable savings in simulation modelling and computation expenses.
Keywords: deep neural network, building simulation, Energia Design Synthesis
Acknowledgement
First Author gratefully acknowledges receiving funding from the Bilateral Scientific and Technological
(TÉT) Cooperation grant program name (#Grant-2019-2.1.11-TÉT) and the support of the Szentágothai
János Research Centre.
HEALTHY AND SUSTAINABLE RENOVATION WITH RENEWABLE MATERIALS
39
Tamás Storcz 1, Kristóf Roland Horváth 2, István Kistelegdi 3, Zsolt Ercsey 4
1. Department of Systems and Software Technologies, University of Pécs, Faculty of Engineering and Information Technology, 7624
Pécs, Boszorkány street 2, storcz.tamas@mik.pte.hu
2. Marcel Breuer Doctoral School, University of Pécs, Faculty of Engineering and Information Technology, 7624 Pécs, Boszorkány
street 2, horvath.kristof.roland@mik.pte.hu
3. Energy Design Research Group, University of Pécs, Faculty of Engineering and Information Technology, 7624 Pécs, Boszorkány
street 2, kistelegdisoma@mik.pte.hu
4. Department of Systems and Software Technologies, University of Pécs, Faculty of Engineering and Information Technology, 7624
Pécs, Boszorkány street 2, ercsey.zsolt@pte.hu
Energia Design Synthesis (EDS) is a method to design buildings considering both energy efficiency while
offering the best comfort. Earlier as part of the method, based on geometry, energy and climate-related
rules, all potentially optimal building configurations were generated, and complex energy and comfort
simulations were performed to identify the optimal design solution for some simple classes of problems.
For the investigated simple classes, serious limiting architectural rules had to be considered to evaluate
each potentially feasible situation.
To broaden the range of the covered cases, the EDS method is extended, and some artificial intelligence-
based techniques are applied. Here, based on the dataset of some previous complex dynamic thermal
simulations performed, an artificial neural network is built and optimised. On the elaborated dataset,
precision and recall are determined. With the help of supervised learning, further cases with not yet
investigated input parameters can be predicted and, in addition, future optimization can be carried out
with considerable savings in simulation modelling and computation expenses.
Keywords: deep neural network, building simulation, Energia Design Synthesis
Acknowledgement
First Author gratefully acknowledges receiving funding from the Bilateral Scientific and Technological
(TÉT) Cooperation grant program name (#Grant-2019-2.1.11-TÉT) and the support of the Szentágothai
János Research Centre.
HEALTHY AND SUSTAINABLE RENOVATION WITH RENEWABLE MATERIALS
39
