Predicting stress–strain relationships of additively manufactured materials under compression and tension using transfer learning and Wasserstein distance-based dataset pruning

• A transfer learning model predicts stress–strain curves of additively manufactured materials. • Four distances measure the distance between data distributions in the source and target domains. • A dataset pruning approach significantly reduces the amount of required training data. Additive manufacturing allows one to design materials and structures with tunable mechanical properties. However, predicting the stress–strain relationships of additively manufactured materials remains a challenge due to complex process-structure–property relationships in additive manufacturing. While machine learning has been used to predict stress–strain relationships, it usually requires large volumes of training data, which is time consuming and expensive. To address this issue, a generic transfer learning (TL) framework that integrates distance-based dataset pruning (DP) is developed. Four distance metrics, including Euclidean, Cosine, Mahalanobis, and Wasserstein distances, are used to prune redundant data in the source domain by determining the distance between the source domain and the training dataset in the target domain. Only the most representative stress–strain curves in the source domain are retained to pre-train a long short-term memory model. We demonstrate the TL framework on two datasets, including fused filament fabrication (FFF) fabricated polylactic acid (PLA) compressive dataset and laser powder bed fusion (L-PBF) fabricated AlSi10Mg tensile dataset. Experimental results show that TL integrated with Wasserstein distance-based DP achieves the best predictive performance and computational efficiency, with the average mean absolute percentage error of 15.18 % for Case 1 and 19.51 % for Case 2.