About this Abstract |
Meeting |
MS&T21: Materials Science & Technology
|
Symposium
|
Additive Manufacturing: Large-Scale Metal Additive Manufacturing
|
Presentation Title |
A Machine Learning-based Geometric Compensation Method for Metal Additive Manufacturing |
Author(s) |
Wen Dong, Albert C. To |
On-Site Speaker (Planned) |
Wen Dong |
Abstract Scope |
The large thermal gradients induced during the metal additive manufacturing (AM) lead to residual stress and deformation in the as-built part and the build plate, thereby degrading the performance and quality of the product and increasing the difficulty of post-processing like machining and cutting. The present work develops a machine learning-based geometric compensation method to reduce the distortion caused by metal AM processes. The method includes three steps: (1) employ both numerical simulations and experimental optical scanning to determine the distortion of the as-built part; (2) establish the mapping relation between the designed and the as-built geometry by neural networks, and generate the compensated shape; (3) convert the compensated shape to CAD files that manufacturers can directly use. The experimental validation shows that the proposed approach is able to effectively improve the geometric accuracy of the as-built part and reduce the residual stress level. |