TY - JOUR
T1 - Universal machine learning for topology optimization
AU - Chi, Heng
AU - Zhang, Yuyu
AU - Tang, Tsz Ling Elaine
AU - Mirabella, Lucia
AU - Dalloro, Livio
AU - Song, Le
AU - Paulino, Glaucio H.
N1 - Funding Information:
The authors acknowledge the financial support from Siemens Corporate Technology under the project titled “Deep Learning Enhanced Topology Optimization”. The inception of this research was the Siemens FutureMakers Challenge at Georgia Tech (May 4-5, 2018), which included an institute-wide hackathon. Two authors of this paper were members of the hackathon winning team: HC, YZ & GHP. The other members of the hackathon team were Mrs. Emily D. Sanders and Mr. Yang Jiang. GHP was the hackathon faculty advisor for the team. HC and GHP acknowledge the support from the Raymond Allen Jones Chair at the Georgia Institute of Technology .
Publisher Copyright:
© 2019 Elsevier B.V.
PY - 2021/3/1
Y1 - 2021/3/1
N2 - We put forward a general machine learning-based topology optimization framework, which greatly accelerates the design process of large-scale problems, without sacrifice in accuracy. The proposed framework has three distinguishing features. First, a novel online training concept is established using data from earlier iterations of the topology optimization process. Thus, the training is done during, rather than before, the topology optimization. Second, a tailored two-scale topology optimization formulation is adopted, which introduces a localized online training strategy. This training strategy can improve both the scalability and accuracy of the proposed framework. Third, an online updating scheme is synergistically incorporated, which continuously improves the prediction accuracy of the machine learning models by providing new data generated from actual physical simulations. Through numerical investigations and design examples, we demonstrate that the aforementioned framework is highly scalable and can efficiently handle design problems with a wide range of discretization levels, different load and boundary conditions, and various design considerations (e.g., the presence of non-designable regions).
AB - We put forward a general machine learning-based topology optimization framework, which greatly accelerates the design process of large-scale problems, without sacrifice in accuracy. The proposed framework has three distinguishing features. First, a novel online training concept is established using data from earlier iterations of the topology optimization process. Thus, the training is done during, rather than before, the topology optimization. Second, a tailored two-scale topology optimization formulation is adopted, which introduces a localized online training strategy. This training strategy can improve both the scalability and accuracy of the proposed framework. Third, an online updating scheme is synergistically incorporated, which continuously improves the prediction accuracy of the machine learning models by providing new data generated from actual physical simulations. Through numerical investigations and design examples, we demonstrate that the aforementioned framework is highly scalable and can efficiently handle design problems with a wide range of discretization levels, different load and boundary conditions, and various design considerations (e.g., the presence of non-designable regions).
UR - http://www.scopus.com/inward/record.url?scp=85098683108&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85098683108&partnerID=8YFLogxK
U2 - 10.1016/j.cma.2019.112739
DO - 10.1016/j.cma.2019.112739
M3 - Article
AN - SCOPUS:85098683108
SN - 0045-7825
VL - 375
JO - Computer Methods in Applied Mechanics and Engineering
JF - Computer Methods in Applied Mechanics and Engineering
M1 - 112739
ER -