Component Tolerance Assessment Using Testing and Machine Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for assessing the quality of manufactured gas turbine engine components rely solely on manual inspection of dimensions, which can lead to misclassification of components as unacceptable when they are actually functionally tolerant.

Innovation Solution

A dual branch assessment system combining physical testing and machine learning is employed, where data from multiple locations on the component is analyzed through a testing branch and a machine learning branch, ensuring both methods agree on the component's functional tolerance before acceptance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual inspection of dimensions is used to assess component quality, then the assessment process is simple to implement, but it leads to misclassification of components as unacceptable when they are actually functionally tolerant

Engineering Contradiction:
Improveassessment accuracyVSAvoidassessment system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The assessment system is segmented into two independent branches: a testing branch that performs traditional dimensional tolerance checks and a machine learning branch that evaluates functional tolerance. Each branch operates independently and contributes to the final assessment decision, allowing the system to leverage both traditional and advanced methods without requiring complete system redesign

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent combines the testing branch and machine learning branch into a unified dual-branch assessment system. Both branches process the same component data but apply different assessment criteria, and their results are integrated to make the final acceptance/rejection decision, thereby improving overall assessment accuracy while maintaining manageable system complexity

Inventive Principle:
Principle #5Merging (Combining)

2Productivity

If manual inspection methods are used for component assessment, then the implementation is straightforward, but the evaluation speed is slow and misclassification occurs

Engineering Contradiction:
Improveevaluation speedVSAvoidassessment accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent replaces manual inspection methods with an automated machine learning-based assessment system. The machine learning branch processes component data computationally to determine functional tolerance, significantly increasing evaluation speed while improving accuracy by identifying functionally tolerant components that manual inspection would incorrectly reject

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If traditional dimensional tolerance checking is used, then the assessment criteria are clear and objective, but components within functional tolerance but outside nominal tolerance ranges are rejected

Engineering Contradiction:
Improvetolerance assessment flexibilityVSAvoidcomponent acceptance reliability
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The patent introduces a second assessment criterion (functional tolerance) alongside the traditional dimensional tolerance. The machine learning branch evaluates components based on functional tolerance parameters that are less stringent than nominal dimensional tolerances, allowing components with minor dimensional variations to be accepted if they remain functionally tolerant, thereby improving acceptance reliability

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4614264A1Part assessment with testing and machine learning branches
Publication Date: 2025.09.10 PRATT & WHITNEY CANADA CORP
  • EP4614264A1 patent drawingFigure 1
  • EP4614264A1 patent drawingFigure 2
  • EP4614264A1 patent drawingFigure 3

AI summary

A method includes the steps of manufacturing and generating data at each of a plurality of locations on a component (22). Passing the data to a testing branch (25) which performs tests (28, 30) and reaches a conclusion as to whether the component (22) is functionally tolerant at each of the plurality of locations. The data is also passed to a machine learning branch (35) wherein it is compared to training data to determine whether the component (22) is of a functionally tolerant dimension at each of the plurality of locations. The manufactured component accepts should both the testing branch (25) and the machine learning branch (35) determine the component (22) is of functionally tolerant dimensions at the plurality of locations, and rejects the component (22) if either of the testing branch (25) or the machine learning branch (35) determines the component (22) fails to be functionally tolerant dimensions, respectively. A system (20) is also disclosed.