Advancing quality prediction in polymer PBF-LB: a hybrid AI and physics-guided approach

The future of advanced manufacturing relies extensively on process information, multiphysics modeling and systems networking. This study explores a novel approach to part quality prediction combining in-process sensor data from three identical systems, physics-guided model and offline CT measurement. The laser powder bed fusion (PBF-LB) process for polymers served as case technology platform for this study. Data from physics-guided model were utilized to train an AI model capable, after training, to confidently predicts the dimensions of printed artifacts. This work drives a paradigm shift in industrial decision-making, reducing reliance on costly, time-intensive domain-specific expertise.