Ensure Quality of Metal Additive Manufacturing with the help of an AI Solution
duration | 2 hours
Description
This course delivers a practical, proven guide to implement AI solutions that minimize defects in metal additive manufacturing. It begins with an analysis of different types of defects in Powder Bed Fusion (PBF) and Direct Energy Deposition (DED). It then covers camera setup and hardware and finally explores the training and deployment of an AI solution.
Learning Outcomes
By the end of this course, participants will be able to understand how to:
- Deploy a full hardware setup, including high-resolution cameras and light-controlled enclosures (inspection kiosk), to capture standardized and repeatable images of printed parts.
- Plan and integrate such systems into the manufacturing workflow.
- Train and deploy advanced AI models, such as convolutional neural networks (CNNs), to automatically detect common surface-level defects (e.g. discoloration, cracks, pores).
- Test the developed models and systems in a real relevant environment
Introduction to metal additive manufacturing and defect characterization
duration | 1 hours
Learning Outcomes
- Understand the different types of metal additive manufacturing.
- Identify quality indicators for metal additive manufacturing.
- Consider visual properties to monitor for quality assurance of part.
Activities
- Interactive lecture
- Group discussion: “How to select the correct type of monitoring for metal manufacturing process”.
- Case example of metal manufacturing process.
Hardware setup selection and preparation
duration | 0.5 hours
Learning Outcomes
- Learn differences of different camera and lens types.
- Decide on different combinations of camera and lens based on application requirements.
- Learn how to set up a high-resolution camera system and acquire data for model training.
Activities
- Lecture on vision-based sensors
- Group Exercise: Equipment selection for different applications
- Quiz
AI model creation and testing
duration | 1 hours
Learning Outcomes
- Set up a software development environment
- Select appropriate vision models for training
- Decide about pre-processing of acquired data
- Run model training on a split data set
- Validate the trained model on a test data set
Activities
- Exercise: installation and set-up of programming environment
- Lecture on model selection and tuning
- Exercise: execute training and validation