ROLLERinspect developed by Confovis is a holistic defect detection and classification system for rotationally symmetrical bodies. In the past, defects (e.g., dents) in the production of rolling elements were detected visually by combined brightfield and darkfield illumination or by fast line scan cameras and parameterizable evaluation software (e.g., Neurocheck).
This is a qualitative inspection method – however, no numerical information about the defect height is available. The optical measuring system ROLLERinspect, on the other hand, receives defect position data from the defect detection system and transfers the data to the measuring system by means of angle and x-, y- coordinate transformation.
The system moves to defect locations according to size automatically and makes 3D measurements. This provides the user with quantitative information about the defect size, which is related to the tolerance.
Scratches and corrosion
Volume and function parameters
Advantages of defect detection and measurement with ROLLERinspect
- Fast defect detection and classification by proven Neurocheck defect identification
- Transfer of defect coordinates to Confovis by coordinate transformation
- Measurement of the classified defects using confocal microscopy
- Automated evaluation of defects according to defect catalog
- Interface with the company’s database
- Roughness measurements traceable to standards possible
- One-button operation of the measuring system is suitable for the production floor
- User-friendly, intuitive measurement software from Confovis
- Additional evaluation by established MountainsMap© analysis software
ROLLERinspect in application
Quantitative information within seconds
Since defect detection and defect measurement are fully integrated for the first time, there are various advantages for the user. With the possibility of immediate quantification of defects, the defect detection classifier can be set in such a way that neither pseudo defects are detected nor “slippage” in the form of undetected defects occurs. Securing the classifier setting generates a deep learning effect that can be transferred to other defect detection systems on the production floor.
The teaching-in of new test specimens is an intuitive menu-guided process; so, no prior knowledge is required to operate the ROLLERinspect measuring system. Test programs once generated can be called up time and again and the changeover of the hardware to another test specimen is completed in less than a minute with sensor-assisted menu navigation. The measurement results are stored in the company database as numerical values (e.g., heights of points of impact) and as a report with graphical representation of the defect and are available at any time. The measuring system can be integrated without restriction into Industry 4.0 processes.