| DOI | Resolve DOI: https://doi.org/10.1109/CCECE59415.2024.10667108 |
|---|
| Author | Search for: Nazari, Ali; Search for: Aghajani, Armin; Search for: Buhr, Phiona; Search for: Park, Byoungyoul1ORCID identifier: https://orcid.org/0000-0003-4493-8343; Search for: Belov, Miroslav1ORCID identifier: https://orcid.org/0000-0002-6097-7126; Search for: Wang, Yunli2ORCID identifier: https://orcid.org/0000-0002-2320-954X; Search for: Shafai, Cyrus |
|---|
| Affiliation | - National Research Council Canada. Quantum and Nanotechnologies
- National Research Council Canada. Digital Technologies
|
|---|
| Format | Text, Article |
|---|
| Conference | 2024 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE), August 6-9, 2024, Kingston, Ontario, Canada |
|---|
| Subject | MEMS; Lorentz force actuator; multi-objective optimization; surrogate model; Gaussian process regression |
|---|
| Abstract | Micro-electromechanical systems (MEMS) sensors and actuators are widely used in a variety of applications, from medical imaging to space telecommunications, making their optimal design crucial. Designing MEMS is a time-consuming process that requires numerous iterations of resource-intensive simulations to evaluate potential designs. As the number of design variables and objectives grows, the complexity and required computational time for this process also increase significantly. Consequently, most efforts to tackle this challenge have focused on scenarios with limited design parameters and a single objective, leaving the area of efficient multi-objective optimization (MOO) for MEMS devices relatively unexplored. In this study, we employ surrogate-assisted design optimization for a MEMS Lorentz force actuator. During an iterative multi-objective optimization process, surrogate models are utilized for performance evaluation of designs instead of numerical simulations. This approach enables us to achieve optimal designs that satisfy all objective constraints using as low as 2% of the number of simulations required compared to case surrogate models are not used, greatly facilitating design optimization. Additionally, we investigate how the number of training simulations and their preprocessing impact the accuracy of the surrogate models and the optimization results. |
|---|
| Publication date | 2024-09-24 |
|---|
| Publisher | IEEE |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 58c66dcb-321f-4ba5-a2f0-a2917dc056e8 |
|---|
| Record created | 2024-11-15 |
|---|
| Record modified | 2024-11-15 |
|---|
