OptiSense - Enhancing Quality and Sustainability with Optical Sensors and Physics-Informed Neural Networks

Purpose and goal

The objective of the proposed project is to develop a system to test the main hypothesis that Physics-Informed Neural Networks (PINNs), using process knowledge combined with surface data (images, surface texture), together with real time and historical process data, can be used to optimize the production of pots and pans with high quality surface finish.

Expected effects and result

The results will be an evaluation of the accuracy of the predictions as well as the potential impact of implementing such control system regarding sustainability including material usage, energy consumption, and other relevant resources. The project will increase the competence and ability of Swedish industry to apply and make use of advanced digitalisation methods and increase the interest in future possibilities in this technology area.

Planned approach and implementation

The project consortium will set up a system with sensors, pattern recognition, database with training data for PINNs from previous production and simulations and train the PINNs. The system will be tested through comparisons with a real running production process.