Supervised pearlitic–ferritic steel microstructure segmentation by U-Net convolutional neural network

The aim of this work is to develop an automated procedure based on machine learning capabilities for the identification of the pearlite islands within the two-phase pearlitic–ferritic steel. The input parameters for the custom implementation of a braided neural network are provided as a data set of...

Full description

Saved in:
Bibliographic Details
Published in:Archives of Civil and Mechanical Engineering 2022-09, Vol.22 (4), p.206, Article 206
Main Authors: Motyl, Mateusz, Madej, Łukasz
Format: Article
Language:English
Subjects:
Artificial neural networks
Automation
Civil Engineering
Data compression
Deep learning
Electron backscatter diffraction
Engineering
Feature extraction
Ferritic stainless steels
Image acquisition
Image reconstruction
Investigations
Learning
Machine learning
Mechanical Engineering
Mechanical properties
Metadata
Microstructure
Morphology
Neural networks
Original Article
Parameters
Pearlite
Scanning electron microscopy
Software quality
Steel
Structural Materials
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The aim of this work is to develop an automated procedure based on machine learning capabilities for the identification of the pearlite islands within the two-phase pearlitic–ferritic steel. The input parameters for the custom implementation of a braided neural network are provided as a data set of scanning electron microscopy images of metallographic specimens. The procedures related to the processing of the data and the optimization parameters affecting the final architecture and effectiveness of the network learning stage are examined. The objective is to find the best solution to the problem of ferritic–pearlitic microstructure segmentation, allowing further processing during, e.g., 3D reconstruction of data from serial sectioning. The work examines the various quality of input data and different U-Net architectures to find the one that can identify pearlite islands with the highest precision. Two types of images acquired from secondary electron (SE) and electron backscattered diffraction (EBSD) detectors are used during the investigation. The work revealed that the developed approach offers improvements in metallographic investigations by removing the requirement for expert knowledge for the interpretation of image data prior to further characterization. It has also been proven that artificial neural networks based on the deep learning process using extensible U-Net network architectures and nonlinear learning tools can identify pearlite islands within a two-phase microstructure, while the overtraining level remains low. Convolutional neural networks do not require manual feature extraction and are able to automatically find appropriate search functions to recognize pearlite structure areas in the training process without human intervention. It was shown that the network recognizes areas of analyzed steel with satisfactory precision of 79% for EBSD and 87% for SE images.
ISSN:2083-3318
1644-9665
2083-3318
DOI:10.1007/s43452-022-00531-4