DefectZero – Advanced Detection and Prevention of Surface Defects
Sheet metal stamping is a cost-effective and widely utilised manufacturing process across numerous industries. However, maintaining consistent quality remains challenging due to variability in material properties and process conditions, often resulting in surface defects and significant resource waste. Machine vision and AI-driven systems, offer promising solutions for real-time, automated inspection and defect classification. Integrating Physics-Informed Neural Networks (PINNs) with production data and simulations could enable predictive modelling of defects, facilitating adaptive process control and optimisation. This approach holds substantial potential to reduce scrap, improve surface quality, and enhance production efficiency through intelligent, data-driven manufacturing systems. The proposed feasibility study will identify and evaluate potential solutions with a systemic approach and be the basis for a follow-up development and implementation project with the highest impact.