WELD quality evaluation with AI and Digitalization (WELD-AID)

Purpose and goal:

The purpose of the WELD-AID project is to transform the quality assurance of welded structures by integrating artificial intelligence and digitalization. The goal is to strengthen Sweden´s position as a global leader in the design and manufacturing of welded products. The project focuses on improving both the predictive accuracy of the lifespan of welded structures and the ability to classify welds and imperfections using AI.

Expected results and effects:

The WELD-AID project will enable automated inspection and classification of various welds, providing direct feedback on fatigue quality during quality inspection. Additionally, it will offer insights into how process parameters impact the final quality. By utilizing advanced neural networks, the welding process will be optimized, improving predictive accuracy and efficiency in weld quality assurance, and enhancing Sweden´s global industrial competitiveness.

Approach and implementation:

The WELD-AID project aims to enhance the quality assurance of welded structures through AI-based solutions in three areas: developing methods for detecting joints and imperfections using deep learning and data augmentation, improving fatigue life predictions of welds through digital twins and AI, and developing an autonomous welding process control system that combines reinforcement learning with traditional optimization techniques.