GenYacht: An interactive generative design system for computer-aided yacht hull design

Creative Commons License

Khan S., Günpınar E., ŞENER B.

OCEAN ENGINEERING, cilt.191, 2019 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 191
  • Basım Tarihi: 2019
  • Doi Numarası: 10.1016/j.oceaneng.2019.106462
  • Dergi Adı: OCEAN ENGINEERING
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Anahtar Kelimeler: Generative design, Interactive design, Computer-aided design, Yacht hull design, LEARNING-BASED OPTIMIZATION, EVOLUTIONARY DESIGN, GENETIC ALGORITHM, PERFORMANCE, GEOMETRY
  • Yıldız Teknik Üniversitesi Adresli: Evet

Özet

In the present work, a new digital design system, GenYacht, is proposed for the creation of optimal and user-centred yacht hull forms. GenYacht is a hybrid system involving generative and interactive design approaches, which enables users to create a variety of design alternatives. Among them, a user can select a hull design with desirable characteristics based on its appearance and hydrostatics/hydrodynamic performance. GenYacht first explores a given design space using a generative design technique (GDT), which creates uniformly distributed designs satisfying the given design constraints. These designs are then presented to a user and single or multiple designs are selected based on the user's requirements. Afterwards, based on the selections, the design space is refined using a novel space-shrinking technique (SST). In each interaction, SST shrinks the design space, which is then fed into GDT to create new designs in the shrank space for the next interaction. This shrinkage of design space guides the exploration process and focuses the computational efforts on user-preferred regions. The interactive and generative design steps are repeated until the user reaches a satisfactory design(s). The efficiency of GenYacht is demonstrated via experimental and user studies and its performance is compared with interactive genetic algorithms.