Loading...

Topology Optimization to Generative Design for Additive Manufacturing Based on Constructive Solid Geometry and Machine Learning

Lotfi, Mohaddeseh | 2023

39 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56426 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Khodaygan, Saeed
  7. Abstract:
  8. The unique ability of additive manufacturing in manufacturing industrial parts with complex geometries has led to the development of topology optimization methods and their use for the optimal design of parts in various industries. However, considering the limitations of additive manufacturing methods in the optimal design of parts is usually a difficult process, and when making decisions, facing multiple conflicting design goals is often difficult and usually involves high computational cost. The main goal of this research is to provide a topology optimization method for generative design for additive manufacturing based on the concept of constructive solid geometry and using machine learning techniques so that the designer can easily check the manufacturability of the part and take into account the construction constraints in the stage of optimizing the topology and choose the most suitable one. In this method, first the topology’s geometries with specific boundaries and parameteres are produced by using the constructive solid geometry. Then, using genetic optimization algorithm, optimal solutions are searched. In the following, a set of optimal optimal solutions are presented in the form of Pareto front for the designer's decision making. Then, the designer's optimal topology is selected. Next, in order to post-process the optimal topology, using the autoencoder network, sharp corners with stress concentration are corrected. The algorithm of the proposed method is implemented on two different problems and the optimization results are presented in single-objective and multi-objective cases. The Pareto front obtained from the optimization is compared with the related results and is acceptable. The post-processing process applied on the topology in order to reduce the stress concentration in the sharp corners has been successful and the stress distribution has been compared before and after the process. In the review of the results, the effect of the parameters of the problem and autoencoder network has been investigated and the algorithm has been implemented in the most optimal possible mode. In order to show the capability of the proposed algorithm, some case examples are solved using the proposed method and the results are compared and discussed with the existing methods in the background of the research
  9. Keywords:
  10. Additive Manufacturing ; Machine Learning ; Topology Optimization ; Genetic Algorithm ; Generative Design ; Constructive Solid Geometry

 Digital Object List

 Bookmark

No TOC