Loading...

Topology optimization for manufacturability of Additive Manufacturing based on Deep Learning and Generative Adversarial Network

Mohseni, Maedeh | 2023

60 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56562 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Khodaygan, Saeed
  7. Abstract:
  8. In recent years, Additive Manufacturing has been extensively used by various industries. These manufacturing processes produce components in a layer-by-layer manner; therefore, they do not impose any geometric constrains to engineers and provide designers with the freedom to design components. Nowadays, one of the primary goals of all industries is to utilize as few raw materials as possible; this way they can deal with the shortage of raw materials and improve their efficiency. Consequently, they implement topology optimization algorithms to design and produce their components. However, topology optimization algorithms result in complicated geometries that can only be fabricated by AM. Although AM processes are capable of producing complex parts, engineers must consider factors when design for AM in order to achieve the best results without mechanical defects. In this work, we aim to develop intelligent networks to design for Selective Laser Melting and improve its manufacturability for topology optimized structures. In order to improve manufacturability by modifying geometries, we took three critical factors into account and developed Residual Autoencoders and Transfer Learning. Furthermore, we classified building direction with the aim of reducing support materials by training Conditional Generative Adversarial Networks and Residual Networks. The Autoencoder accurately learned the input and output features and generated geometries that are manufacturable by SLM. In addition, Transfer Learning increased the generalization of networks and resulted in an intelligent system to enhance the manufacturability of complex topologies. Finally, training of CGAN and ResNets led to classification of building directions with 99% accuracy
  9. Keywords:
  10. Additive Manufacturing ; Topology Optimization ; Deep Learning ; Transfer Learning ; Generative Adversarial Networks ; Design for Additive Manufacturing

 Digital Object List

 Bookmark

No TOC