Experimental analysis of mechanical properties of biomaterial FDM 3D printed output according to raster direction and implementation of machine learning model

The printing conditions applied in FDM 3D printing are a key factor affecting the mechanical performance of the part. Quantitative analysis of the change in mechanical properties of the printed parts depending on the raster angle during the printing conditions can be used to optimize the FDM 3D printing process. In this study, in order to optimize the FDM 3D printing process of biomaterials using PLA filaments, tensile tests were conducted on the printed parts based on the raster direction to analyze the change of mechanical properties according to the raster angle. The tensile test dataset and the laminated tissue image dataset were input to the machine learning models, LSTM model and FM model, to perform machine learning. In this study, the RNN-based LSTM model was used to create a stress-strain comparison map that closely approximates the tensile test results, and the CNN-based FM model was used to create a feature map that enhances the image pattern of the laminated tissue to analyze the connection between the shape features and mechanical properties of the laminated tissue. The results of this study, which analyzed the changes in mechanical properties of biomaterial FDM3D printing printouts depending on the raster angle, predicted mechanical properties derived from machine learning models, and identified the linkage between mechanical properties by visualizing geometric features, are expected to play a key role in process optimization and establishing the reliability of mechanical performance analysis methodologies in the field of biomaterial FDM 3D printing technology.