Applying Machine Learning for Real-time Structural Reinforcement

Applying Machine Learning for Real-time Structural Reinforcement


Smart Geometry 2018: Machine Minds

AI Strategies for Space Frame Structures Cluster

Space frame structure designed and constructed during the workshops



This project aims to explore a collaborative design process between human and computer intelligence by using Machine Learning methods and finite element analysis.
The project represents an approach to structural optimization which makes use of local rules applied to subsections of larger, complex models. In order to speed up the optimization, an artificial neural network is trained. This approach utilizes test set-ups which comprise a large subset of the locally possible geometric and loading conditions. The viability of this procedure is tested using three geometries which derive from the division of given volumes into arbitrary cells. A comparison of the maximum displacement under dead weight before and after the optimization shows an improvement by a factor of roughly ten. During the smart Geometry Conference in 2018 one of the structures was subsequently built using wire connected by gluing tape.



Cluster Champions:
Zeynep Aksöz(Innochain/OpenFields_research//studio)
Clemens Preisinger(Karamba3d)

Smart Geometry Collaborators:
Erika Stadnik
Natalie Sham
Edward Bruun
Peter Olendzki
Louis Leblanc
Omri Menashe
Xiaolong Li
Michael Lee
Anthony Mattacchione
Andre Pereira
Feng Le