Automatic Identification and Quantitative Characterization of Primary Dendrite Microstructure Based on Machine Learning

Abstract

Dendrites are important microstructures in single-crystal superalloys. The distribution of dendrites is closely related to the heat treatment process and mechanical properties of single-crystal superalloys. The primary dendrite arm spacing (PDAS) is an important length scale to describe the distribution of dendrites. In this work, the second-generation single crystal superalloy HT901 with a diameter of 15 mm was imaged under a metallurgical microscope. An automatic dendrite core identification and full-field quantitative statistical analysis method is proposed to automatically detect the dendrite core and calculate the local PDAS. The Faster R-CNN algorithm combined with test time augmentation (TTA) technology is used to automatically identify the dendrite cores. The local multi-directional algorithm combined with Voronoi tessellation is used to determine the local nearest neighbor dendrite and calculate the local PDAS and coordination number. The accuracy of using Faster R-CNN combined with TTA to detect the dendrite core of HT901 reaches 98.4%, which is 15.9% higher than using Faster R-CNN alone. The algorithm calculates the local PDAS of all dendrites in H901 and captures the Gaussian distribution of the local PDAS. The average PDAS determined by the Gaussian distribution is 415 μm, which is only a small difference from the average spacing λ¯ (420 μm) calculated by the traditional method. The technology analyzes the relationship between the local PDAS and the distance from the center of the sample. The local PDAS near the center of HT901 are larger than those near the edge. The results suggests that the method enables the rapid, accurate and quantitative dendritic distribution characterization.

Correlation among Composition, Microstructure and Hardness of 7xxx Aluminum Alloy Using Original Statistical Spatial-Mapping Method

“…This process requires significant time and labor inputs; thus, the statistical results are acquired only from limited features and regions [22][23][24]. In view of the above shortcomings, in this study, image acquisition instruments were combined with computer vision methods, and a high throughput method of quantitative identification and statistical analysis of the aluminum alloy microstructure was devised based on deep learning [25,26]. After verification, this method was found to be accurate and comprehensive for the statistics of massive organizational data; please refer to our published articles, for the specific research process and results [27].…”

Estimation of Average Grain Size from Microstructure Image Using a Convolutional Neural Network

“…Quantitative microstructure analysis has also been used to interpret alloy components [ 10 ], material properties [ 11 , 12 ], and microstructural features [ 13 , 14 ] based on microstructure images using machine learning and computer vision. From the point of view of machine learning for image recognition, the prediction of alloy components or material properties and the prediction of average grain size look similar to each other, but there is one major difference.…”

Quantitative Characterization of the γ’ Phase Distribution in the Large-Scale Area of the Second-Generation Nickel-Based Single Crystal Blade DD5

“…This work uses the semantic segmentation network U-Net [22] based on deep learning [23,24] to quantitatively and statistically characterize the distribution of γ' phases in a large area of different transverse sections and different parts of the second-generation single crystal blade DD5. The use of deep learning-based methods to achieve the quantitative statistical analysis of the microstructure has been widely used in a variety of materials [25][26][27]. The U-Net network structure uses a completely symmetrical encoding-decoding structure, and uses skip connections to fuse the feature information of each layer in the encoding process into the corresponding layer feature in the decoding process.…”