Non-destructive Quality Testing in Rim Casting
Non-destructive testing (NDT) during rim casting is a critical component of quality assurance in the production of aluminum or magnesium rims for vehicles. NDT methods enable the detection of material defects such as cracks, porosity, blowholes (tiny cavities), or inclusions in the cast material without compromising the integrity of the component.
CBCT (Cone Beam-Computertomografie) image quality varies with the number of the X-ray images taken. In order to speed up scan times in production, a compromise is often made and scan times are reduced by taking fewer images. However, this reduces the image quality of the CBCT reconstructions and thus increases the demand for high-quality AI models. These are then specifically tuned to the image quality and also learn to deal with more CBCT reconstruction artifacts.
The 3D CBCT data obtained with this method offers a rapid and non-destructive comprehensive view of the manufactured casting, providing valuable insight into injection molding defects.
This forms the basis for a detailed 3D analysis with precise defect localization and characterization through exact interpretation of the results, which requires specialist knowledge and experience.
Our expertise lies in the following areas:
- 3D modelling with fully 3D context: our unique selling point
- Training 3D AI models: our core competency
- Rapid adaptation of AI models through augmentation and CBCT simulation
- Analysis of large volumes (512³ or larger) of data in a few seconds
- Optimal performance for a wide range of image quality and image artefacts
In order to achieve 100% inline inspections with minimal production times and the associated short cycle times for CBCT data generation, it is essential that the CBCT analysis functions reliably even with correspondingly limited image quality.
We address this with machine learning functionalities.
Our development expertise in CBCT reconstruction, utilizing reconstruction algorithms such as FDK (Feldkamp) or iterative methods, enables us to train our 3D AI models even on a limited number of training samples. By leveraging real-time data analysis and machine learning, we are able to promptly identify material defects and structural anomalies.
To ensure the most precise outcomes, we create additional training data through the simulation of a genuine reconstruction, incorporating typical anomalies from the CAD model of a component, such as an injection-molded rim.
We integrate our 3D AI model into embedded systems such as the NVIDIA Jetson product family with minimal development effort, offering customized solutions at the highest level. This results in the following advantages for you in the production process.
Customer benefits:
- Seamless integration into existing production processes
- Faster CBCT analyses even with limited image quality
- Improved efficiency in non-destructive quality testing
- Cost reduction in the ongoing operation of CBCT inspection systems in foundries
- Rapid market launch
We are your partner for efficient embedded solutions in the NDT sector. Please contact us to discuss your requirements.
Marcus
Dr.-Ing. Marcus Prümmer
Am Weichselgarten 7
91058 Erlangen
Germany
+49 (0)9131 - 691 385
+49 (0)9131 - 691 386
pruemmer(at)chimaera.de