내진 성능을 고려한 비정형 강재 댐퍼의 딥러닝 기반 생성형 설계

초록

The seismic design of steel dampers is critical for enhancing the structural resilience of buildings under seismic loads. However, achieving cost-effective and tailored solutions remains challenging due to the diverse seismic demands of different structures. This study introduces a generative design framework for unstructured steel dampers, optimizing seismic performance and construction costs. Pareto-optimal solutions derived through optimization form the training dataset for a deep learning generative model, which integrates Variational Autoencoders (VAE) to improve data distribution and ensure feasible designs. This research presents a scalable approach to seismic design, leveraging advanced deep learning techniques and optimization to achieve resilience and cost-efficiency in steel damper applications.

키워드