Methods and apparatus for machine learning predictions of manufacture processes
The subject technology is related to methods and apparatus for discretization and manufacturability analysis of computer assisted design models. In one embodiment, the subject technology implements a computer-based method for the reception of an electronic file with a digital model representative of a physical object. The computer-based method determines geometric and physical attributes from a discretized version of the digital model, a cloud point version of the digital model, and symbolic functions generated through evolutionary algorithms. A set of predictive machine learning models is utilized to infer predictions related to the manufacture process of the physical object.
1. A system for predictions of manufacturing processes, comprising:
a graphical user interface tool including computerized instructions for receiving, at a compute device, an electronic file including a digital model representative of a physical object to be manufactured;
an object analysis engine including computerized instructions for generating, from the digital model, a first set of parameters associated with the physical object;
a first machine learning engine including computerized instructions for generating a second set of parameters associated with the physical object from at least the first set of parameters; and
a second machine learning engine including computerized instructions for producing at least one predictive outcome associated with manufacturing the physical object, based at least in part upon the second set of parameters, and based at least in part upon one or more neural network machine learning models;
wherein at least one of the one or more neural network machine learning models is a regression model;
wherein the at least one predictive outcome corresponds to a manufacturing process; and
wherein the graphical user interface tool further includes computerized instructions for providing to a user responsive information associated with manufacturing the physical object, the responsive information being based upon the at least one predictive outcome.
2. The system of claim 1 , wherein the first set of parameters include at least one of the following: (i) values corresponding to non-planar surface areas associated with the physical object; (ii) values corresponding to a volume of the physical object, (iii) values corresponding to a surface area of the physical object, (iv) values corresponding to convex hull volume of the physical object, and (v) values corresponding to length, width and height of a prism enclosing the physical object.
3. The system of claim 1 , wherein at least one of the neural network machine learning models is trained using unsupervised machine learning.
4. The system of claim 1 , wherein the object analysis engine comprises a mesh analysis engine.
5. The system of claim 1 , wherein the object analysis engine comprises a point cloud analysis engine.
6. The system of claim 1 , wherein the object analysis engine comprises a mesh analysis engine and a point cloud analysis engine.
7. The system of claim 1 ,
wherein the first machine learning engine includes a symbolic function engine including instructions for determining, from at least the first set of parameters, at least one of functional forms or approximations of targeted functions describing at least one of physical attributes or geometric attributes of a physical object.
8. The system of claim 7 , wherein the symbolic function engine utilizes an evolutionary computation model trained with samples including a representative collection of digital models of physical objects.
9. The system of claim 1 , wherein the second machine learning engine produces the at least one predictive outcome based further upon one or more candidate shape distributions retrieved as a pre-classifier of physical object shape.
10. The system of claim 1 , wherein the second machine learning engine produces a plurality of predictive outcomes taken from a group consisting of: price, set-up time, cycle time, a number of Computer Numerical Control (CNC) operations, requirement of a mill machine, requirement of a lathe machine, requirement of a sheet metal process, a type of blank used, or a type of fixture used.
11. The system of claim 1 , wherein the system provides the responsive information to the graphical user interface tool in near real-time upon receiving the electronic file from the graphical user interface tool.
12. A system for predictions of manufacturing processes, comprising:
a graphical user interface tool including computerized instructions for receiving, at a compute device, an electronic file including a digital model representative of a physical object to be manufactured, and for returning responsive information associated with manufacturing the physical object, the responsive information including information related at least to a predictive outcome generated by a predictive engine;
wherein the predictive outcome generated by the predictive engine is generated by the predictive engine based at least in part upon (a) at least one of a set of parameters associated with the digital model or information derived from the set of parameters, and (b) one or more neural network machine learning models, at least one of the one or more neural network machine learning models being trained using a collection of digital models of a domain of mechanical objects;
wherein the at least one predictive outcome corresponds to one or more of the following: predicted cost, set-up time, cycle time, a number of Computer Numerical Control (CNC) operations, requirement of a mill machine, requirement of a lathe machine, requirement of a sheet metal process, a type of blank used, or a type of fixture used;
wherein the graphical user interface also receives an identification of a fabrication material; and
at least one of the one or more neural network machine learning models predicts whether or not it is feasible to manufacture the physical object using the identified fabrication material.
13. The system of claim 12 , wherein the set of parameters is generated from the digital model by an object analysis engine based upon at least one of (i) a discretized version of the digital model or (ii) a point cloud processing of the digital model.
14. The system of claim 12 , wherein the set of parameters include (i) values corresponding to surface orientations associated with the physical object and (ii) values corresponding to non-planar surface areas associated with the physical object.
15. The system of claim 14 , wherein the set of parameters further include a plurality of the following: (iii) values corresponding to a volume of the physical object, (iv) values corresponding to a surface area of the physical object, (v) values corresponding to convex hull volume of the physical object, and (vi) values corresponding to length, width and height of a prism enclosing the physical object.
16. The system of claim 12 , wherein the predictive engine produces a plurality of predictive outcomes taken from a group consisting of: set-up time, cycle time, a number of Computer Numerical Control (CNC) operations, requirement of a mill machine, requirement of a lathe machine, requirement of a sheet metal process, a type of blank used, or a type of fixture used.
17. The system of claim 12 , wherein at least one of the one or more neural network machine learning models is trained using unsupervised machine learning.