Machine learning technique to identify grains in polycrystalline materials samples
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.
1. A method of identifying grains in polycrystalline materials, the method comprising:
(a) identifying local crystal structure of the polycrystalline material based on neighbor coordination or pattern recognition machine learning, the local crystal structure comprising 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.
2. The method of claim 1 , wherein the step of identifying local crystal structure is based on neighbor coordination and comprises identifying the atomic structure of a first neighbor of the grains and grain boundaries as at least one of hexagonal close packing (hcp), face-centered cubic (fcc), body-centered cubic (bcc), and icosahedral.
3. The method of claim 2 , wherein the step of identifying local crystal structure further comprises identifying the atomic structure of a second neighbor of the grains and grain boundaries as at least one of hexagonal close packing (hcp), face-centered cubic (fcc), body-centered cubic (bcc), and icosahedral.
4. The method of claim 3 , wherein the step of identifying local crystal structure generates voxels and a number count of:
(a) each type of atomic structure for the first neighbor; and
(b) each type of atomic structure for the second neighbor.
5. The method of claim 1 , wherein the step of identifying local crystal structure is based on unsupervised machine learning.
6. The method of claim 4 , wherein the step of pre-processing comprises applying a uniform filter to the voxels to reduce noise within the grains and improve contrast of the grain boundaries.
7. The method of claim 4 , further comprising labeling the voxels as either in the grain or at the grain boundary.
8. The method of claim 7 , wherein the step of conducting grain identification comprises segregating individual grains by classifying the voxels based on grain index and assigning the voxels to be a portion of the grain.
9. The method of claim 8 , wherein the step of refining the grains comprises reassigning voxels labeled as at the boundary to its spatially nearest grain.