Machine learning in tolerancing for additive manufacturing

Design for additive manufacturing has gained extensive research attention in recent years, whereas tolerancing issues aiming at controlling geometric variations remain a major bottleneck in achieving predictive models and realistic simulations. In this paper, a prescriptive deviation modelling method coupled with machine learning techniques is proposed to address the modelling of shape deviations in additive manufacturing. The in-plane geometric deviations are mapped into an established deviation space and Bayesian inference is used to estimate geometric deviations patterns by statistical learning from multiple shapes data. The effectiveness of the proposed approach is demonstrated and discussed through illustrative case studies.