# US 11,079,738 B2 · Framework for rapid additive design with generative techniques

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 11,079,738 B2
- Document ID: US11079738B2
- Status: Granted
- Kind: B2
- Application number: 15/678,653
- Filed: 2017-08-16
- Published: 2019-02-21
- Granted: 2021-08-03
- Earliest claimed: 2017-08-16
- Assignees: General Electric Company
- Applicants: General Electric Company
- Inventors: Arun Karthi Subramaniyan; Ananda Barua; Daniel Erno
- Web page: https://patentgrep.com/patent/US11079738B2

## CPC

- G05B 19/4099: Surface or curve machining, making three-dimensional [3D] objects, e.g. desktop manufacturing
- G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G06F 30/17: Mechanical parametric or variational design
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/0475: Generative networks
- G06N 3/09: Supervised learning
- G06N 3/094: Adversarial learning
- G06N 5/025: Extracting rules from data
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- B33Y 50/00: Data acquisition or data processing for additive manufacturing
- G05B 2219/49023: 3-D printing, layer of powder, add drops of binder in layer, new powder
- G06N 3/08: Learning methods
- G06N 5/027: {Frames}
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- B33Y: ADDITIVE MANUFACTURING, i.e. MANUFACTURING OF THREE-DIMENSIONAL [3D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVE AGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3D PRINTING, STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING

## Abstract

According to some embodiments, a system may include a design experience data store containing electronic records associated with prior industrial asset item designs. A deep learning model platform, coupled to the design experience data store, may include a communication port to receive constraint and load information from a designer device. The deep learning platform may further include a computer processor adapted to automatically and generatively create boundaries and geometries, using a deep learning model associated with an additive manufacturing process, for an industrial asset item based on the prior industrial asset item designs and the received constraint and load information. According to some embodiments, the deep learning model computer processor is further to receive design adjustments from the designer device. The received design adjustments might be for example, used to execute an optimization process and/or be fed back to continually re-train the deep learning model.

## Claims

### Claim 1

A system to facilitate creation of an industrial asset item, comprising:
a design experience data store containing electronic records associated with prior industrial asset item designs; and
a deep learning model platform, coupled to the design experience data store, including:
a communication port to receive constraint and load information from a designer device, and
a deep learning model computer processor coupled to the communication port and adapted to automatically and generatively create boundaries and geometries, using a deep learning model associated with an additive manufacturing process, for the industrial asset item based on the prior industrial asset item designs and the received constraint and load information, wherein the automatically and generatively created boundaries and geometries are associated with a generative model for the industrial asset item.

### Claim 2 (depends on 1)

The system of claim 1, further comprising:
a deep learning model creation platform to receive the electronic records associated with the prior industrial asset item designs and to automatically create the deep learning model using three-dimensional images of multiple industrial asset item designs.

### Claim 3 (depends on 1)

The system of claim 1, wherein the deep learning model computer processor is further to receive design adjustments from the designer device and, based on the received design adjustments, execute an optimization process.

### Claim 4 (depends on 3)

The system of claim 3, wherein the received design adjustments are fed back to continually re-train the deep learning model.

### Claim 5 (depends on 1)

The system of claim 1, further comprising:
a physics model platform to receive the boundaries and geometries and execute a validation process on at least one intermediate industrial asset design.

### Claim 6 (depends on 5)

The system of claim 5, wherein results of the validation process are fed back to continually re-train the deep learning model.

### Claim 7 (depends on 1)

The system of claim 1, further comprising:
a repository storing a plurality of searchable generative models.

### Claim 8 (depends on 7)

The system of claim 7, wherein a designer can perform an automated search of the repository.

### Claim 9 (depends on 8)

The system of claim 8, wherein results of said search comprise generative model recommendations ranked by probabilities.

### Claim 10 (depends on 1)

The system of claim 1, wherein the generative model is created by a Generative Adversarial Network (“GAN”) based on at least one contextual model or at least one image model.

### Claim 11 (depends on 1)

The system of claim 1, wherein the generative model is associated with at least one of: (i) a Gaussian mixture model, (ii) a hidden Markov model, (iii) probabilistic context-free grammar, (iv) a naive Bayes model, (v) averaged one-dependence estimators, (vi) latent Dirichlet allocation, (vii) a restricted Boltzmann machine, and (xiii) a generative adversarial network.

### Claim 12 (depends on 1)

The system of claim 1, wherein the generative model is further created based on natural language processing of at least one of: (i) text, (ii) reports, (iii) design handbooks, and (iv) design practices.

### Claim 13 (depends on 12)

The system of claim 12, wherein the natural language processing creates a contextual model.

### Claim 14 (depends on 1)

The system of claim 1, wherein the generative model is further created based on classification and feature identification performed by a deep neural network in connection with at least one of: (i) an image, (ii) a manufactured design, (iii) a cross-section, (iv) a binary Computer Aided Design (“CAD”) file, (v) a geometry file, and (vi) analysis results.

### Claim 15 (depends on 14)

The system of claim 14, wherein the classification and feature identification creates an image model.

### Claim 16 (depends on 1)

The system of claim 1, further comprising:
an additive manufacturing printer, to receive a final industrial asset design based on the automatically and generatively created boundaries and geometries, wherein the printer associated with at least one of: (i) three-dimensional printing, (ii) vat photopolymerization, (iii) material jetting, (iv) binder jetting, (vi) material extrusion, (vii) powder bed fusion, (viii) sheet lamination, and (ix) directed energy deposition.

### Claim 17

A computer-implemented method to facilitate creation of an industrial asset item, comprising:
receiving, at a search platform from a designer device, a search of a generative model repository;
executing a search of the generative model repository to identify a starting seed shape for the industrial asset item;
receiving, at a deep learning model platform, the starting seed shape along with constraint and load information from the designer device;
automatically and generatively creating boundaries and geometries, by the deep learning model platform using a deep learning model associated with an additive manufacturing process, for the industrial asset item based on the prior industrial asset item designs and the received constraint and load information;
receiving design adjustments at the deep learning model platform from the designer device;
executing an optimization process based on the received design adjustments, execute an optimization process;
executing, at a physics model platform, a validation process on at least one intermediate industrial asset design based on received the boundaries and geometries;
receiving, at an additive manufacturing printer, a final industrial asset design based on the automatically and generatively created boundaries and geometries; and
creating, by the additive manufacturing printer, the industrial asset item.

### Claim 18 (depends on 17)

The method of claim 17, wherein the automatically and generatively created boundaries and geometries are associated with a generative model for the industrial asset item.

### Claim 19 (depends on 18)

The method of claim 18, wherein the generative model is created by a Generative Adversarial Network (“GAN”) based on at least one contextual model or at least one image model.

### Claim 20 (depends on 19)

The method of claim 19, wherein the generative model is associated with at least one of: (i) a Gaussian mixture model, (ii) a hidden Markov model, (iii) probabilistic context-free grammar, (iv) a naive Bayes model, (v) averaged one-dependence estimators, (vi) latent Dirichlet allocation, (vii) a restricted Boltzmann machine, and (xiii) a generative adversarial network.

### Claim 21 (depends on 20)

The method of claim 20, wherein the generative model is further created based on a contextual model produced via natural language processing of at least one of: (i) text, (ii) reports, (iii) design handbooks, and (iv) design practices.

### Claim 22 (depends on 20)

The method of claim 20, wherein the generative model is further created based on an image model produced via classification and feature identification performed by a deep neural network in connection with at least one of: (i) an image, (ii) a manufactured design, (iii) a cross-section, (iv) a binary Computer Aided Design (“CAD”) file, (v) a geometry file, and (vi) analysis results.