Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning

Research output: Contribution to journalArticleResearchpeer review

Authors

Research Organisations

External Research Organisations

  • Tongji University
  • Ho Chi Minh City University of Technology (HUTECH)
View graph of relations

Details

Original languageEnglish
Pages (from-to)516-535
Number of pages20
JournalFrontiers of Structural and Civil Engineering
Volume18
Issue number4
Publication statusPublished - Apr 2024

Abstract

Identifying crack and predicting crack propagation are critical processes for the risk assessment of engineering structures. Most traditional approaches to crack modeling are faced with issues of high computational costs and excessive computing time. To address this issue, we explore the potential of deep learning (DL) to increase the efficiency of crack detection and forecasting crack growth. However, there is no single algorithm that can fit all data sets well or can apply in all cases since specific tasks vary. In the paper, we present DL models for identifying cracks, especially on concrete surface images, and for predicting crack propagation. Firstly, SegNet and U-Net networks are used to identify concrete cracks. Stochastic gradient descent (SGD) and adaptive moment estimation (Adam) algorithms are applied to minimize loss function during iterations. Secondly, time series algorithms including gated recurrent unit (GRU) and long short-term memory (LSTM) are used to predict crack propagation. The experimental findings indicate that the U-Net is more robust and efficient than the SegNet for identifying crack segmentation and achieves the most outstanding results. For evaluation of crack propagation, GRU and LSTM are used as DL models and results show good agreement with the experimental data.

Keywords

    crack propagation, crack segmentation, deep learning, encoder-decoder, recurrent neural network

ASJC Scopus subject areas

Cite this

Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning. / Tran, Than V.; Nguyen-Xuan, H.; Zhuang, Xiaoying.
In: Frontiers of Structural and Civil Engineering, Vol. 18, No. 4, 04.2024, p. 516-535.

Research output: Contribution to journalArticleResearchpeer review

Tran, TV, Nguyen-Xuan, H & Zhuang, X 2024, 'Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning', Frontiers of Structural and Civil Engineering, vol. 18, no. 4, pp. 516-535. https://doi.org/10.1007/s11709-024-1040-z
Tran, T. V., Nguyen-Xuan, H., & Zhuang, X. (2024). Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning. Frontiers of Structural and Civil Engineering, 18(4), 516-535. https://doi.org/10.1007/s11709-024-1040-z
Tran TV, Nguyen-Xuan H, Zhuang X. Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning. Frontiers of Structural and Civil Engineering. 2024 Apr;18(4):516-535. doi: 10.1007/s11709-024-1040-z
Tran, Than V. ; Nguyen-Xuan, H. ; Zhuang, Xiaoying. / Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning. In: Frontiers of Structural and Civil Engineering. 2024 ; Vol. 18, No. 4. pp. 516-535.
Download
@article{4bd25f592672469099b74aa06f9ac8fb,
title = "Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning",
abstract = "Identifying crack and predicting crack propagation are critical processes for the risk assessment of engineering structures. Most traditional approaches to crack modeling are faced with issues of high computational costs and excessive computing time. To address this issue, we explore the potential of deep learning (DL) to increase the efficiency of crack detection and forecasting crack growth. However, there is no single algorithm that can fit all data sets well or can apply in all cases since specific tasks vary. In the paper, we present DL models for identifying cracks, especially on concrete surface images, and for predicting crack propagation. Firstly, SegNet and U-Net networks are used to identify concrete cracks. Stochastic gradient descent (SGD) and adaptive moment estimation (Adam) algorithms are applied to minimize loss function during iterations. Secondly, time series algorithms including gated recurrent unit (GRU) and long short-term memory (LSTM) are used to predict crack propagation. The experimental findings indicate that the U-Net is more robust and efficient than the SegNet for identifying crack segmentation and achieves the most outstanding results. For evaluation of crack propagation, GRU and LSTM are used as DL models and results show good agreement with the experimental data.",
keywords = "crack propagation, crack segmentation, deep learning, encoder-decoder, recurrent neural network",
author = "Tran, {Than V.} and H. Nguyen-Xuan and Xiaoying Zhuang",
note = "Publisher Copyright: {\textcopyright} The Author(s) 2024.",
year = "2024",
month = apr,
doi = "10.1007/s11709-024-1040-z",
language = "English",
volume = "18",
pages = "516--535",
number = "4",

}

Download

TY - JOUR

T1 - Investigation of crack segmentation and fast evaluation of crack propagation, based on deep learning

AU - Tran, Than V.

AU - Nguyen-Xuan, H.

AU - Zhuang, Xiaoying

N1 - Publisher Copyright: © The Author(s) 2024.

PY - 2024/4

Y1 - 2024/4

N2 - Identifying crack and predicting crack propagation are critical processes for the risk assessment of engineering structures. Most traditional approaches to crack modeling are faced with issues of high computational costs and excessive computing time. To address this issue, we explore the potential of deep learning (DL) to increase the efficiency of crack detection and forecasting crack growth. However, there is no single algorithm that can fit all data sets well or can apply in all cases since specific tasks vary. In the paper, we present DL models for identifying cracks, especially on concrete surface images, and for predicting crack propagation. Firstly, SegNet and U-Net networks are used to identify concrete cracks. Stochastic gradient descent (SGD) and adaptive moment estimation (Adam) algorithms are applied to minimize loss function during iterations. Secondly, time series algorithms including gated recurrent unit (GRU) and long short-term memory (LSTM) are used to predict crack propagation. The experimental findings indicate that the U-Net is more robust and efficient than the SegNet for identifying crack segmentation and achieves the most outstanding results. For evaluation of crack propagation, GRU and LSTM are used as DL models and results show good agreement with the experimental data.

AB - Identifying crack and predicting crack propagation are critical processes for the risk assessment of engineering structures. Most traditional approaches to crack modeling are faced with issues of high computational costs and excessive computing time. To address this issue, we explore the potential of deep learning (DL) to increase the efficiency of crack detection and forecasting crack growth. However, there is no single algorithm that can fit all data sets well or can apply in all cases since specific tasks vary. In the paper, we present DL models for identifying cracks, especially on concrete surface images, and for predicting crack propagation. Firstly, SegNet and U-Net networks are used to identify concrete cracks. Stochastic gradient descent (SGD) and adaptive moment estimation (Adam) algorithms are applied to minimize loss function during iterations. Secondly, time series algorithms including gated recurrent unit (GRU) and long short-term memory (LSTM) are used to predict crack propagation. The experimental findings indicate that the U-Net is more robust and efficient than the SegNet for identifying crack segmentation and achieves the most outstanding results. For evaluation of crack propagation, GRU and LSTM are used as DL models and results show good agreement with the experimental data.

KW - crack propagation

KW - crack segmentation

KW - deep learning

KW - encoder-decoder

KW - recurrent neural network

UR - http://www.scopus.com/inward/record.url?scp=85194703385&partnerID=8YFLogxK

U2 - 10.1007/s11709-024-1040-z

DO - 10.1007/s11709-024-1040-z

M3 - Article

AN - SCOPUS:85194703385

VL - 18

SP - 516

EP - 535

JO - Frontiers of Structural and Civil Engineering

JF - Frontiers of Structural and Civil Engineering

SN - 2095-2430

IS - 4

ER -