| Abstract Scope |
Design for Additive Manufacturing (DfAM) extends traditional design for manufacturing principles to additive manufacturing (AM), enabling the optimization of design and manufacturing processes for layer-by-layer fabrication. Implicit modeling represents complex geometries as function fields, enabling efficient processing of intricate shapes and diverse manufacturing data compared to conventional CAD-based methods. However, implicit modeling poses challenges for applying traditional feature extraction techniques, limiting effective use of geometric information necessary for AM optimization. This study proposes methodology for extracting AM features based on implicit modeling, along with a corresponding optimization framework for AM processes. Convolutional neural network extracts AM features from implicit field data, which are then reintegrated into the implicit model as scalar fields. An optimization method is applied according to each AM feature’s characteristics, improving overall AM process performance. The framework successfully extracted five AM features (tower, thin wall, hole, bridge, overhang) from evaluation datasets, significantly enhancing print quality through optimization. |