期刊
COMPUTATIONAL MATERIALS SCIENCE
卷 160, 期 -, 页码 186-196出版社
ELSEVIER SCIENCE BV
DOI: 10.1016/j.commatsci.2019.01.006
关键词
Microstructure classification; Data mining; Support vector machine (SVM); Haralick image texture; Morphological parameter; SEM; LOM; Low-carbon low-alloy steel
The variety of modern steels is growing steadily. In order to meet the ever tighter tolerance ranges for the properties of these steels, it is important to both understand the manufacturing process as accurately as possible and to be able to correctly classify the microstructure. The microstructure acts as a link between the production and the properties, which acts as an information storage, which must be read out and understood in the best way to develop new steels. For this reason, it is of utmost importance to have an objective and reproducible microstructure classification available. The present study demonstrates that using a support vector machine in combination with pixel-based and morphology-based parameters allows a reliable classification based on microstructural images. In order to determine the parameters correlative microscopy is used to collect a large variety of information about the different steel structures. The significance of the different parameter groups for the classification success and the correlation of the parameters with each other are investigated. The minimum number of parameters required for a reliable classification is determined by evolutionary feature selection.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据