Machine learning for sensorless induction welding of carbon fiber thermoplastics
| Reference number | |
| Coordinator | Corebon Production AB |
| Funding from Vinnova | SEK 500 000 |
| Project duration | December 2020 - August 2021 |
| Status | Completed |
| Venture | AI - Competence, ability and application |
| Call | Start your AI-journey for businesses - autumn 2020 |
Important results from the project
The project investigates the use of AI methods to interpret data from an induction heating system, such as voltage, current, frequency and phase angle to predict the temperature of the material being heated, otherwise difficult to measure. The application is induction welding of carbon fiber thermoplastic composite for joining products of lightweight material within e.g. aerospace, automotive and wind power industries. The results are very promising, and commercialization expected in 2022-2023.
Expected long term effects
A high-risk project with uncertainty about what could be achieved, but the results from the AI model are great, and reliable temperature prediction can be achieved despite variations in material quality, thickness, etc. The technology enables reliable welding of components that is currently not possible on an industrial scale and has great potential to become a de-facto standard in the industry. It remains to implement the algorithms in the equipment to be able to be used for real-time control.
Approach and implementation
The project has shown that AI models have enormous potential to not only predict valuable but difficult-to-measure quantities from various data, but also to quantify its reliability or accuracy. However, the amount and quality of data used to train the model is critical, as is the choice of model. Implementing this type of algorithm on a built-in system for real-time use is entirely possible, but places high demands on both model and hardware.