Manufacturing inspection systems and associated methods

Patent metadata

Field	Value
Publication number	US-2026072420-A1
Application number	US-202418883077-A
Country	US
Kind code	A1
Filing date	Sep 12, 2024
Priority date	Sep 12, 2024
Publication date	Mar 12, 2026
Grant date	—

Abstract

A system for inspecting a gas turbine engine component includes an inspection environment operable to access feature information associated with a component design. The feature information includes a unique identifier assigned to a respective geometric feature of the component design. The environment is operable to access one or more inspection criterion associated with the unique identifier. The environment is operable to access process information obtained during manufacturing of a physical instance of the respective geometric feature according to one or more manufacturing parameters. The environment is operable to generate, using a machine learning model, a prediction of whether the physical instance of the respective geometric feature meets the one or more inspection criterion based on the process information. A method for inspecting a component is also disclosed.

First claim

What is claimed is:

1 . A system for inspecting a gas turbine engine component comprising:
one or more processors coupled to memory, the one or more processors collectively operable to execute an inspection environment, and the inspection environment operable to: access feature information associated with a component design, the feature information including a unique identifier assigned to a respective geometric feature of the component design; access one or more inspection criterion associated with the unique identifier; access process information obtained during manufacturing of a physical instance of the respective geometric feature according to one or more manufacturing parameters; generate, using a machine learning model, a prediction of whether the physical instance of the respective geometric feature meets the one or more inspection criterion based on the process information; and generate an indicator associated with the prediction.

2 . The system as recited in claim 1 , wherein the feature information includes at least one of the following: a geometry of the geometric feature, a dimension of the geometric feature, and a tolerance associated with the dimension.

3 . The system as recited in claim 2 , wherein the one or more inspection criterion includes at least one of the following: the tolerance associated with the dimension; and
one or more manufacturing acceptance criterion.

4 . The system as recited in claim 1 , wherein the inspection environment operable to:
cause the one or more manufacturing parameters to be adjusted based on the prediction; and generate, using the machine learning model, a prediction of whether another physical instance of the respective geometric feature manufactured according to the adjusted one or more manufacturing parameters meets the one or more inspection criterion.

5 . The system as recited in claim 1 , wherein the inspection environment is operable to:
access one or more manufacturing repositories including a manufacturing execution repository, wherein:
entries in the one or more manufacturing repositories are associated with unique identifiers assigned to respective geometric features of one or more component designs to establish a set of digital threads linking the respective entries across the one or more manufacturing repositories by the respective unique identifier;
the manufacturing execution repository includes one or more instructions to manufacture the respective geometric features of the one or more component designs; and
the one or more manufacturing parameters include the respective one or more instructions associated with manufacture of the physical instance of the respective geometric feature; and
identify an entry in the manufacturing execution repository assigned the unique identifier including the respective one or more instructions utilized to manufacture the physical instance of the respective geometric feature, wherein the prediction is based on the identified entry.

6 . The system as recited in claim 1 , wherein the machine learning model is operable to generate the prediction without any physical inspection information corresponding to the physical instance of the geometric feature.

7 . The system as recited in claim 1 , wherein the machine learning model includes a neural network.

8 . The system as recited in claim 1 , wherein:
training data utilized to train the machine learning model includes the prediction associated with a prior physical instance of manufacturing the geometric feature.

9 . The system as recited in claim 1 , wherein:
training data utilized to train the machine learning model includes inspection information obtained during physical inspection of the prior instance of manufacturing the geometric feature.

10 . The system as recited in claim 1 , wherein the component design is associated with a gas turbine engine component.

11 . A non-transitory computer-readable medium having computer-executable instructions that, when executed by one or more processors, cause the one or more processors to collectively execute an inspection environment operable to:
access feature information associated with a component design, the feature information including a unique identifier assigned to a respective geometric feature of the component design; access one or more inspection criterion associated with the unique identifier; access process information obtained during manufacturing of a physical instance of the respective geometric feature according to one or more manufacturing parameters; generate, using a machine learning model, a prediction of whether the physical instance of the respective geometric feature meets the one or more inspection criterion based on the process information; and generate an indicator associated with the prediction.

12 . The non-transitory computer-readable medium as recited in claim 11 , wherein:
the feature information includes at least one of the following: a geometry of the geometric feature, a dimension of the geometric feature, and a tolerance associated with the dimension; the one or more inspection criterion includes the tolerance associated with the dimension; and the indicator is associated with the tolerance.

13 . The non-transitory computer-readable medium as recited in claim 11 , the inspection environment is operable to:
cause the one or more manufacturing parameters to be adjusted based on the prediction; and generate, using the machine learning model, a prediction of whether another physical instance of the respective geometric feature manufactured according to the adjusted one or more manufacturing parameters meets the one or more inspection criterion.

14 . The non-transitory computer-readable medium as recited in claim 11 , wherein the machine learning model includes a neural network.

15 . A method for inspecting a component comprising:
accessing feature information associated with a component design including a geometric feature associated with a respective unique identifier; accessing one or more inspection criterion associated with the unique identifier; manufacturing a physical instance of the respective geometric feature of a component according to one or more manufacturing parameters; accessing process information obtained during the manufacturing that is associated with the unique identifier; generating, using machine learning, a prediction of whether the physical instance of the respective geometric feature meets the one or more inspection criterion based on the process information and/or the one or more manufacturing parameters; and accepting or rejecting the physical instance of the component based on the prediction.

16 . The method as recited in claim 15 , wherein the one or more inspection criterion include a dimension and/or a tolerance associated with the unique identifier.

17 . The method as recited in claim 15 , further comprising:
accessing one or more manufacturing repositories including a manufacturing execution repository, wherein entries in the one or more manufacturing repositories are associated with unique identifiers assigned to respective geometric features of one or more component designs to establish a set of digital threads linking the respective entries across the one or more manufacturing repositories by the respective unique identifier; determining the digital thread associated with the physical instance of the respective geometric feature; and identifying an entry in the manufacturing execution repository associated with the determined digital thread, the e

Assignees

• Rtx Corp

Inventors

• Guo Changsheng
• Vega Albino Hermitt

Classifications

• G05B2219/32368

  Quality control · CPC title

• G06N20/10

  using kernel methods, e.g. support vector machines [SVM] · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N20/20

  Ensemble learning · CPC title

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title