Skip to main content
. 2024 Apr 23;37:348–377. doi: 10.1016/j.bioactmat.2024.03.033

Fig. 11.

Fig. 11

A machine learning-evolutionary algorithm (ML-EA) to assist the design of 4D printed active composites with desired shape transformation. (a) Property mismatch-induced actuation and distribution of two materials with different shape memory properties (property “1” and property “2” are encoded as “1” and “0” for the algorithm, respectively). (b) A complete design for designing and optimizing the 4D printing path, including generation of the dataset by FEA simulations of random designs, prediction of shape changes by ML, optimization of material distribution by EA, and 4D printing of active composite using the optimized design. (c) 4D printing process based on the ML-EA design, including identification of drawn profiles as target shapes, ML-EA design, conversion of the obtained optimal design to the grayscale slices, 4D printing by DLP, and actuation of the printed structure. (d) Experimental results showing DLP-produced structures with target shape transformation to mimic drawn profiles. Reproduced with permission [159]. Copyright 2021, Wiley-VCH.