Grain Identification in Polycrystalline Materials Using 3D Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current grain size analysis technologies are limited to two-dimensional data sets and cannot effectively identify grains in three-dimensional polycrystalline materials, which is a critical limitation for understanding the properties and applications of these materials in technological and energy contexts.
Innovation Solution
A method involving machine learning techniques, specifically using neighbor coordination and unsupervised machine learning, to identify local crystal structures and refine grain boundaries in polycrystalline materials, including pre-processing with image processing to improve noise reduction and contrast, and segregating voxels to classify grains, is employed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional 2-D image processing techniques are used for grain analysis, then the analysis can be performed with simple methods, but the technique cannot be extended to 3-D data sets
Solution Approach 1:
The patent extends grain analysis from traditional two-dimensional images to three-dimensional data sets by implementing a 3-D neighborhood coordination analysis method. This allows the system to process volumetric data (such as from atom probe tomography or electron microscopy) and identify grains in three-dimensional space, thereby achieving adaptability to 3-D data sets while maintaining a systematic analytical framework
Solution Approach 2:
The patent introduces machine learning algorithms as an intermediary between raw 3-D data and grain identification. The machine learning model serves as a mediator that automatically learns the complex patterns and features of grain structures in 3-D space, eliminating the need for manual feature engineering and reducing the apparent complexity of the analysis technique
2Measurement precision
If machine learning techniques are implemented for 3-D grain identification, then accurate grain analysis in 3-D data sets is achieved, but computational complexity and processing requirements increase
Solution Approach 1:
The patent segments the 3-D data set into local neighborhoods or voxels, where each neighborhood is analyzed independently for its coordination properties. This segmentation allows the machine learning model to process smaller, manageable units rather than the entire 3-D data set at once, reducing computational complexity while maintaining accurate grain identification through localized analysis
Solution Approach 2:
The patent performs preliminary processing of the 3-D data by calculating neighborhood coordination numbers and structural features before applying the machine learning model. This preliminary action pre-computes essential features that simplify the subsequent grain identification task, reducing the computational burden on the machine learning algorithm while maintaining high accuracy
Data Source
AI summary
A method of identifying grains in polycrystalline materials, the method including (a) identifying local crystal structure of the polycrystalline material based on neighbor coordination or pattern recognition machine learning, the local crystal structure including grains and grain boundaries, (b) pre-processing the grains and the grain boundaries using image processing techniques, (c) conducting grain identification using unsupervised machine learning; and (d) refining a resolution of the grain boundaries.