Anomaly detection in aerosol jet printing via computer vision & machine learning
The Aerosol Jet Printing (AJP) technique is a relatively new contactless direct write method that is being developed for the purpose of producing fine features on a diverse selection of surfaces. The technique was originally designed to make electronic circuits, but it has subsequently been tested f...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/168392 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | The Aerosol Jet Printing (AJP) technique is a relatively new contactless direct write method that is being developed for the purpose of producing fine features on a diverse selection of surfaces. The technique was originally designed to make electronic circuits, but it has subsequently been tested for many uses, which include active and passive electrical components. While it is a great technique, AJP does have its limitations. For instance, there is still no concrete in-situ monitoring system in place for AJP so that anomalies occurring during the printing process can be detected early. Hence, this paper aims to develop, and more importantly, fine-tune, an in-situ monitoring algorithm that will be implemented into the Aerosol Jet Printing system with the help of an Object Detection Model, which hinges upon the principles of Deep Learning (DL), and Computer Vision (CV). This model was trained on a dataset comprising of six different classes, evaluated against several metrics, and finally, fined-tuned based on its hyperparameters to attain optimal performance. |
|---|