Langeveld, Emma (2024) Benchmarking MAG-weld Visual Inspection Systems. Design Project, Industrial Engineering and Management.
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
MasterIEMDesignProject2024ELangeveld.pdf.pdf Restricted to Registered users only Download (23MB) |
Abstract
AWL-Techniek is a system integrator whom develops automated production systems for the automotive industry among others. The R&D team of AWL is developing an automated weld seam inspection solution. This solution contains three elements: product positioning, acquisition and weld analysis. For the weld analysis, deep learning vision models are used. Multiple suppliers can provide these models to execute the weld analysis. In this research an unbiased comparison method is developed in which the solutions from different suppliers are evaluated. The unbiased comparison method contains of 7 different functions each with an individual goal. For each function, metrics are defined using expert input and relevant literature. The deep learning vision models are evaluated on performance, robustness and soft metrics such as integration and costs. The evaluation contains a defined set of steps with defined test data requirements, ensuring that each solution is assessed the equally. Validation is performed by assessing the flow & instructions of the method, its metrics and the performance. It is found that the comparison method is unbiased and objective. The implementation by software developers is executed without any problems, indicating the instructions and flow of the method are valid.
| Item Type: | Thesis (Design Project) |
|---|---|
| Supervisor name: | Kloosterman, H. and Jayawardhana, B. |
| Degree programme: | Industrial Engineering and Management |
| Thesis type: | Design Project |
| Language: | English |
| Date Deposited: | 29 Oct 2024 06:49 |
| Last Modified: | 29 Oct 2024 06:49 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/34350 |
Actions (login required)
 |
View Item |