Meet the COROB Open Call Winners: Interview with SHINEWELD
What is the solution?
SHINEWELD (Smart Spectral Imaging for Welding Quality ) is an AI-powered optical emission spectroscopy system that monitors welding processes in real-time and automatically classifies weld quality with up to 87% accuracy. By analysing the spectral signature of the welding arc, SHINEWELD instantly detects five types of defects (porosity, inclusions, lack of penetration, geometric errors and correct welds), eliminating the need for costly post-production inspection. SHINEWELD integrates seamlessly with manual and robotic welding equipment, providing immediate quality feedback to operators and production systems.
What makes it novel:
SHINEWELD is the first commercial system to combine:
- Optical emission spectroscopy (proven analytical chemistry technique)
- Machine learning classification (Random Forest algorithms)
- Real-time process monitoring (in-process, not post-production)
- Multi-defect detection (5 simultaneous defect categories)
What motivated you to apply for the COROB Open Call?
SMASP had experience with multispectral cameras, and we saw COROB as an opportunity to expand our knowledge using spectroscopy. In previous agricultural projects, we knew that predictions based on spectral signatures had very good results, and we were motivated to expand its use in the industrial field through welding. When we saw COROB’s call, we realised it was the right opportunity to develop these capabilities.
How do you envision your project making an impact?
SHINEWELD within COROB has been trained only for MIG welding and S235 steel welding with suitable filler material and shielding gas. In order for SHINEWELD to be used as a welding validation method, it is necessary to retrain with more materials.
On the other hand, SHINEWELD has achieved greater accuracy than initially anticipated, making it a faster and safer method that can complement or replace some of the methods currently in use.
How do I see SHINEWELD in the future? In large-scale industrial production lines (automobiles, household appliances, etc.), it will be incorporated into welding robots, so that the weld can be validated almost as soon as it is made, resulting in a significant reduction in time and an increase in quality. As I mentioned before, welding with the same material and similar processes are optimal use cases for SHINEWELD. On the other hand, in welding where the material is changed continuously, its application is more complicated.
Learn more about this solution at: https://smasp.org/web1/shineweld/
