The field of additive manufacturing, especially 3D printing, has gained growing attention in the research and commercial sectors in recent years. Notwithstanding that the capabilities of 3D printing have moved on to enhanced resolution, higher deposition rate, and a wide variety of materials, the crucial challenge of verifying that the component manufactured is within the dimensional tolerance as designed continues to exist. Material extrusion 3D printing has long been established for rapid prototyping and functional testing in many research and industry fields. However, its inconsistency and intrinsic defects (surface roughness and geometric inaccuracies) hinder its application in several areas, most notably “certify-as-you- build” small-batch prototyping and large-batch production.
In this study, we present an approach to reduce both inconsistency and the 3D geometric inaccuracies of products fabricated by material extrusion.
1. This work developed and demonstrated an approach for layer-by-layer mapping of 3D printed parts, which can be used for validation of printed models and in situ adjustment of print parameters. This in situ metrology system scans each layer at the time of printing, providing a 3D model of the as-printed part. A high-speed optical scanning system was integrated with a Material Extrusion type 3D printer to achieve in situ monitoring of dimensional inaccuracies during printing, which leaves the door open to implement a closed-loop feedback system to compensate geometric errors during printing in the future and fabricate “certify-as-you-build” products.
2. This work trained machine learning algorithms with data from this scanning system and predicted 3D geometric inaccuracies in new designs. Eight Conditional Adversarial Networks (CAN) machine learning models were trained on a limited number of scanned profile images of different layers, consisting of less than 50 actual images and 50 generated images, to predict the 3D geometric deviations of freeform shapes. The generated images were produced by randomly combining and cropping the actual images without any distortion. These CAN models produced predictions where at least 44.4%, 87.6%, 99.2% of data were within �0.05 mm, �0.10 mm, �0.15 mm of the actual measured value, respectively.
3. This work developed an Iterative Forward approach to redesign the Computer-Aided- Design model by reverse engineering using the trained machine learning models, allowing for compensation of print imperfection at the design stage, in advance of the first printing. The compensation algorithms with eight different sets of different parameters were evaluated. It has been proven that the Iterative Forward approach improved the geometric deviation of the predicted profiles by making compensation to the CAD model.