Options
Presentation of an optimized process for generating analysis models in structural engineering using Graph Neural Networks
Citation Link: https://doi.org/10.15480/882.13522
Publikationstyp
Conference Paper
Date Issued
2024-09-18
Sprache
English
Author(s)
TORE-DOI
Start Page
440
End Page
447
Citation
35. Forum Bauinformatik, fbi 2024: 440-447
Contribution to Conference
Publisher
Technische Universität Hamburg, Institut für Digitales und Autonomes Bauen
Peer Reviewed
true
Building Information Modeling (BIM) methodology is increasingly used in the construction industry. When this method is utilized for the structural integrity analysis of architectural models, these models must be converted into finite element models. However, automated processes of model conversion are prone to errors, as relevant connections between BIM objects are often missing. Hence, structural engineers often prefer the manual creation of the analysis model in practice. The aim of this paper is to introduce new methods for an efficient and collaborative workflow for automated model conversion. To this end, the use of an AI system in the form of a Graph Neural Network (GNN) is proposed. This system aims to recognize the load-bearing components in a BIM model through semantic enrichment and to establish their relationships based on load transfer in the form of a graph model. Incorrect connections or missing links between building elements are identified using a GNN approach to repair these issues and create correct and computationally efficient structural analysis models. Through these optimized processes, the need for individuall model adaption is expected to be reduced in the future.
Subjects
Building Information Modeling (BIM)
Industry Foundation Classes (IFC)
machine learning
structural engineering
DDC Class
720: Architecture
624.1: Structural Engineering
004: Computer Sciences
006: Special computer methods
Loading...
Name
Presentation of an optimized process for generating analysis models in structural engineering using Graph Neural Networks.pdf
Size
270.88 KB
Format
Adobe PDF