Loading...

Multi Objective Topology Design based on Moving Morphable Component and Machine Learning

Ramezani, Mohammad Sajad | 2022

52 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 56436 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Khodaygan, Saeed
  7. Abstract:
  8. Today, the additive manufacturing process is one of the most important manufacturing processes. In the additive manufacturing process, the mass of the parts is directly related to the printing duration, in addition to the costs of consumables. On the other hand, the use of lighter parts in industries, such as the aviation, is one of the important requirements of those industries, and topology optimization is needed to reduce the consumables and the mass of the parts. Also, since in the optimal design of engineering parts, more than one design objective is usually considered, the multi-objective optimization of topology is of great importance. The Moving Morphable Components is one of the new topology optimization methods that uses geometry explicitly in the optimization process; therefore, it is much easier to create a CAD file from the obtained design. In this research, the optimal topology design is conducted in three sections. In the first section, the topology optimization algorithm based on the MMC approach is developed. In this approach, the optimization results are greatly sensitive to the parameters of the moving asymptotic (MMA) optimizer. In order to make it easier to tune these parameters, the globally convergent method (GCMMA) has been used instead of the primary optimizer of the moving asymptote method (MMA) and the results have been compared in different case studies. According to the results, the proposed method reaches the optimal solution of the problem with less sensitivity to the value of the optimization parameters; however, in terms of computational cost, it often 2-3 times more time. In the second section, the multi-objective topology optimization problem has been analyzed. As a basic test, dual objective optimization problem of simultaneous minimization of compliance and material volume is considered. First, the Pareto front is obtained using the non-dominant sorting genetic algorithm (NSGA II) method. The genetic algorithm does not provide an acceptable answer because of the topology generation method in the MMC approach; therefore, using the analytical method of ϵ-constraint and GCMMA as the optimizer, the problem was solved once again and the results are reported accordingly. When the topology optimization process is done, it is usually necessary to process the obtained designs due to the shortcomings of the modeling and optimization algorithm. Most the processes are done manually by the designer. The last section of the research is dedicated to perform automatic post-processing on the optimal topology using machine learning. For this purpose, a Denoising convolutional Autoencoder (DAE) network has been trained and used at the end of the optimization process, with the aim of removing the noise of sharp corners and stress concentration from the optimal structure. The results showed that the proposed approach is effective in eliminating stress concentration effects
  9. Keywords:
  10. Topology Optimization ; Design for Additive Manufacturing ; Autoencoder ; Multiobjective Optimization ; Movable Morphable Components ; Non-Dominate Sorting Genetic Algorithm (NSGAII) Method ; Additive Manufacturing

 Digital Object List

 Bookmark

No TOC