US 12,282,718 B2
Apparatus and method for generating a manufacturability analysis
Austin Perry, Maple Plain, MN (US); Joshua Knutson, Maple Plain, MN (US); Robert Ford, Maple Plain, MN (US); Samuel Bosak, Maple Plain, MN (US); and Shuji Usui, Maple Plain, MN (US)
Assigned to Proto Labs, Inc., Maple Plain, MN (US)
Filed by PROTOLABS, INC., Maple Plain, MN (US)
Filed on Jun. 9, 2023, as Appl. No. 18/208,145.
Prior Publication US 2024/0411963 A1, Dec. 12, 2024
Int. Cl. G06F 30/27 (2020.01); G06F 119/18 (2020.01)
CPC G06F 30/27 (2020.01) [G06F 2119/18 (2020.01)] 18 Claims
OG exemplary drawing
 
1. An apparatus for generating a manufacturability analysis, wherein the apparatus comprises:
at least a graphics processing unit (GPU); and
a memory communicatively connected to the at least a GPU, the memory containing instructions configuring the at least a GPU to:
receive a computer model that is representative of a part for manufacture, wherein the computer model comprises a plurality of model based definitions;
generate a depth buffer model of the part for manufacture, wherein the depth buffer model further comprises a plurality of depth buffers, wherein each depth buffer of the plurality of depth buffers defines a partial surface of the representative part for manufacture;
generate a plurality of orientations based at least on the plurality of depth buffers, wherein generating the plurality of orientations comprises identifying at least an unreachable zone of the depth buffer model for each orientation of the plurality of orientations, wherein the at least unreachable zone is a feature that is required to be machined in each orientation of the plurality of the orientations;
generate a manufacturability analysis as a function of the plurality of the model based definitions using a manufacturability classifier, wherein the manufacturability classifier is trained as a function of manufacturability training data, wherein the manufacturability training data comprises a plurality of data entries, wherein the plurality of data entries each including at least a model-based definition as an input correlated to a manufacturability analysis as an output, and wherein the manufacturability training data is iteratively updated with results of the manufacturability classifier using a feedback loop;
generate an adjusted computer model as a function of the manufacturability analysis and the computer model;
initiate an approval protocol based on the adjusted computer model, wherein the approval protocol includes an alert comprising an identification of existing unmanufacturable features within the adjusted computer model to a user; and
wherein the adjusted computer model is generated using an adjustment machine learning model.