The project aims for zero-defect production by enabling early surface defect detection and prompt control actions. It introduces advanced 3D and spectral band-specific detectors to enhance defect identification. Additionally, data optimisation methods, including unsupervised ASIS domain adaptation and synthetic data generation, improve classification reliability.
Domaine: Manufacturing ⊕
Factsheet:
The SurfConInspect (SCI) project aims to enable zero-defect manufacturing for flat steel production through an early detection of surface defects and a fast and adequate control action once a defect appears. Therefore, the project follows a holistic approach incorporating new concepts as well for the measuring of surface defects as for the support of adequate control actions.
To optimise surface inspection results a 3D detector and a spectral
band specific detector will be applied for a more reliable detection of surface defects. Additionally, it will be evaluated whether methods for unsupervised ASIS domain adaptation and/or synthetic ASIS data generation, complemented by an open-data concept, can help achieve more reliable classification by increasing the amount of high-quality training data.
For the support of adequate control actions, a modular SCI framework will be implemented able to provide in-coil control actions as well for the operator as directly for the process control systems. To support the operator the applicability of Augmented Reality (AR) devices for the online visualisation of quality information directly on the moving coil will be investigated.
CETIC is responsible for setting up the project Testbed, which will be used for integrating and testing the project framework and evaluating the different approaches for their efficiency and accuracy. Our role is to establish the data flows: a) from the steel plant data sources to the project framework for analysis and b) deliver the surface inspection evaluation results back to the plant user interfaces as well as the project’s XR devices.
Additionally, CETIC is preparing an Open Data repository with surface inspection images from the steel plants that will be published following the Open Data principles.
As a research centre in ICT with many years of involvement in research projects within the steel industry, CETIC aims to contribute to yet another successful solution for a challenging problem in steel manufacturing with high accuracy and efficiency. Projects like this allow us to leverage our expertise in ICT while gaining a deeper understanding of the needs and challenges of the steel industry. This, in turn, helps us enhance our technologies and expertise for future projects in the field.
View online : SurfConInspect Project
