K-Fold Part Assessment for Functionally Tolerant Components
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for assessing the quality of manufactured gas turbine engine components, such as integrally bladed rotors, rely solely on geometric tolerances, which often misidentify functionally tolerant parts as unacceptable, leading to inefficiencies in the evaluation process.
Innovation Solution
A dual branch assessment method combining physical testing and machine learning, utilizing K-fold validation, to evaluate components by generating data at multiple locations, comparing it to training data, and determining functional tolerance through a conservative evaluation of likelihood percentages.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual inspection of dimensions is used to assess component quality, then measurement precision is maintained, but productivity is reduced and functionally tolerant parts are misidentified as unacceptable
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated system that combines 3D scanning, mesh generation, and machine learning algorithms. The system automatically compares as-manufactured geometry against nominal geometry and functional tolerance criteria, eliminating manual measurement while maintaining or improving precision through computational analysis.
Solution Approach 2:
The patent transforms the assessment approach by changing from strict geometric tolerance compliance to a functional tolerance evaluation based on performance parameters. The machine learning model assesses whether deviations from nominal dimensions actually affect component function, allowing parts that are functionally tolerant to be correctly identified despite geometric variations.
2Manufacturing precision
If strict geometric tolerance compliance is enforced, then manufacturing precision is maintained, but reliability is reduced by rejecting functionally tolerant components
Solution Approach 1:
The patent fundamentally changes the assessment parameters from purely geometric dimensions to functional performance criteria. The machine learning model evaluates whether dimensional deviations actually impact component function, transitioning from a rigid geometric compliance approach to a nuanced functional tolerance approach that maintains reliability while reducing false rejections.
Solution Approach 2:
The patent introduces machine learning algorithms as an intermediary between geometric measurement and acceptance decisions. This intermediary layer analyzes the relationship between dimensional variations and functional performance, determining whether deviations from nominal geometry actually affect component reliability, thus mediating between manufacturing precision and functional acceptance.
3Reliability
If comprehensive physical testing is performed on all components, then reliability is improved, but loss of time and productivity are worsened
Solution Approach 1:
The patent applies partial action by performing comprehensive functional analysis only when necessary, rather than on every component. The machine learning model quickly assesses most components and identifies only those requiring detailed physical testing, reducing overall testing time while maintaining reliability by focusing resources on cases where functional concerns are detected.
Solution Approach 2:
The patent performs preliminary assessment using the machine learning model before committing to time-consuming physical testing. The system quickly evaluates geometric data and predicts functional tolerance, allowing only components that require further investigation to proceed to comprehensive physical testing, thus reducing overall time loss while maintaining reliability through staged evaluation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of assessing the quality of a manufactured component includes the steps of manufacturing a component (22) and generating data (24) at each of a plurality of locations on the component (22). The generated data is passed to a machine learning branch (35) wherein the generated data is compared to training data at each of the plurality of locations and across a plurality of folds using K-fold validation to determine whether the component (22) is of a functionally tolerant dimension at each of the plurality of sections. A system (20) is also disclosed.