Volume Image PCA for Geometric Dispersion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for quantifying microstructural variability and mechanical property dispersions in composite parts, such as aircraft turbomachinery blades, are costly and inefficient, particularly when using X-ray tomography and volumetric image correlation, as they fail to effectively illustrate dispersions across a series of parts.
Innovation Solution
A method involving volumetric X-ray tomography image processing that includes correlating images to obtain displacement fields, applying dimensionality reduction via principal component analysis (PCA) to eigenmodes, and performing statistical analysis to identify geometric dispersions and anomalies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If volumetric image correlation is applied to quantify geometric dispersion between parts, then measurement precision is improved, but device complexity increases due to the need for advanced image processing algorithms
Solution Approach 1:
The patent introduces displacement fields as an intermediary representation between the raw volumetric images and the final geometric dispersion metrics. By computing displacement fields that map each part to a reference part, the system transforms complex image comparison into a more manageable mathematical representation, reducing the processing complexity while maintaining measurement precision
Solution Approach 2:
The patent transforms the geometric dispersion measurement problem by changing the parameter space from direct image coordinate comparisons to displacement field representations. This parameter transformation allows the use of statistical analysis tools on the displacement fields, simplifying the overall measurement process while preserving accuracy
2Productivity
If dimensional reduction methods are applied to displacement fields, then productivity is improved by enabling efficient statistical analysis, but measurement precision may be compromised through information loss
Solution Approach 1:
The patent extracts the essential geometric variation information from the full displacement fields by identifying and retaining only the dominant modes of variation. Through eigenmode decomposition, the system separates the significant geometric features from the noise and minor variations, enabling efficient statistical analysis while preserving the critical measurement information
Solution Approach 2:
The patent segments the complex displacement field data into distinct eigenmodes, each representing a specific pattern of geometric variation. This segmentation allows the statistical analysis to focus on the most significant modes independently, improving computational efficiency while maintaining measurement accuracy through the preservation of dominant variation patterns
3Ease of operation
If statistical analysis is performed on eigenmodes to identify defective parts, then ease of operation is improved for quality control, but measurement precision requirements increase to accurately detect anomalies
Solution Approach 1:
The patent creates a statistical model (a kind of digital copy) of the expected geometric variations from reference parts. By comparing the eigenmode coefficients of test parts against this statistical model, the system automatically identifies deviations that indicate defects. This copying approach simplifies the quality control operation while maintaining high detection accuracy through statistical comparison rather than manual inspection
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach allows for efficient and cost-effective quantification of geometric dispersions and identification of defective parts by projecting displacement fields onto eigenmode spaces, facilitating the detection of manufacturing anomalies and monitoring production lines.
Implementation Method 1
This experimental method exploits the differential absorption of X-rays by different materials to reconstruct, through computation, a three-dimensional image of the part under study from a series of radiographs
Data Source
Figure 1
Figure 2
AI summary
Disclosed is a method for processing a plurality of x-ray tomography volume images (I_1, …, I_N) each associated with a part, the plurality of volume images including a reference volume image, comprising: • a step (P_CIV) of correlating volume images in order to obtain a displacement field between each image and the reference image, in order to obtain a plurality of displacement fields minimising the difference between the volume images, • processing, by a dimensionality reduction method (P_PCA), the plurality of image displacement fields in order to express them according to eigenmodes, • a statistical analysis of the fields expressed according to the eigenmodes.