Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision

This study aims to investigate the macro-micro deterioration mechanism of graded gravel under cyclic loading, which is of great significance in revealing the generation mechanism of high-speed railway subgrade defects in service and guiding the subgrade reinforcement. Initially, a precise-rapid quan...

Full description

Saved in:
Bibliographic Details
Main Authors: Tai-feng Li, Xian-pu Xiao, Rong-hui Yan, Kang Xie, Jia-shen Li, Ruo-han Dai
Format: Article
Language:English
Published: Elsevier 2024-12-01
Series:Case Studies in Construction Materials
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2214509524011628
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846151855858515968
author Tai-feng Li
Xian-pu Xiao
Rong-hui Yan
Kang Xie
Jia-shen Li
Ruo-han Dai
author_facet Tai-feng Li
Xian-pu Xiao
Rong-hui Yan
Kang Xie
Jia-shen Li
Ruo-han Dai
author_sort Tai-feng Li
collection DOAJ
description This study aims to investigate the macro-micro deterioration mechanism of graded gravel under cyclic loading, which is of great significance in revealing the generation mechanism of high-speed railway subgrade defects in service and guiding the subgrade reinforcement. Initially, a precise-rapid quantification method for particle shape based on machine vision was established. To further improve the existing traditional semantic segmentation model (U-Net), a novel semantic segmentation model named VGG16-UNet-CA was established, which incorporated the U-Net, VGG16 model, Bilinear interpolation, and Coordinate attention (CA). Moreover, the high-precision binary images obtained by the established model were imported into Open Source Computer Vision Library (OpenCV) to calculate the particle shape indicators rapidly. Additionally, the cyclic loading tests were carried out to characterize the macro deterioration of graded gravel based on the mechanical indicator dynamic stiffness K, and the key factor of deterioration was explored through the established machine vision method. Finally, biaxial compression models with different deterioration degree were established by Discrete Element Method (DEM) to explore the micro deterioration mechanism. It was found that the segmentation accuracy indicators F1-score, Mean Pixel Accuracy (MPA) and Mean Intersection over Union (MIoU) of VGG16-UNet-CA were 98.56 %, 97.89 % and 98.01 %, which were higher than U-Net, SegNet, PSPNet, and DeepLabv3+. During the cyclic loading, the changes of the equivalent particle diameter De and slenderness ratio Ei of coarse particles in three filler groups were not significant, while the average values of roundness Rc increased by 28 %, 31 % and 25 %, indicating that the coarse particle abrasion was the key factor of deterioration. In the DEM simulation, with the increase of Rc, the mechanical strength of graded gravel decreased and the micro indicators (e.g., sliding rate, strong force chains, and anisotropy parameter an) inside the fillers gradually decreased, while the particle rotation continuously increased. Consequently, the interlocking ability of coarse particles decreased, leading to the bearing capacity and structural stability of particle skeleton reduced. This study not only provides a novel approach to investigate the deterioration mechanism of graded gravel under cyclic loading, which contributes to revealing the generation mechanism of high-speed railway subgrade defects in service, but also provides theoretical support for subgrade reinforcement.
format Article
id doaj-art-b50d9c3ff47c4d4c897310ce0c856a04
institution Kabale University
issn 2214-5095
language English
publishDate 2024-12-01
publisher Elsevier
record_format Article
series Case Studies in Construction Materials
spelling doaj-art-b50d9c3ff47c4d4c897310ce0c856a042024-11-27T05:02:37ZengElsevierCase Studies in Construction Materials2214-50952024-12-0121e04011Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine visionTai-feng Li0Xian-pu Xiao1Rong-hui Yan2Kang Xie3Jia-shen Li4Ruo-han Dai5Railway Engineering Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, ChinaRailway Engineering Research Institute, China Academy of Railway Sciences Corporation Limited, Beijing 100081, ChinaSchool of Intelligent Manufacturing and Smart Transportation, Suzhou City University, Suzhou, Jiangsu 215104, China; Corresponding authors.School of Intelligent Manufacturing and Smart Transportation, Suzhou City University, Suzhou, Jiangsu 215104, China; Corresponding authors.Department of Civil Engineering, Central South University, Changsha, Hunan 410075, ChinaSchool of Intelligent Manufacturing and Smart Transportation, Suzhou City University, Suzhou, Jiangsu 215104, ChinaThis study aims to investigate the macro-micro deterioration mechanism of graded gravel under cyclic loading, which is of great significance in revealing the generation mechanism of high-speed railway subgrade defects in service and guiding the subgrade reinforcement. Initially, a precise-rapid quantification method for particle shape based on machine vision was established. To further improve the existing traditional semantic segmentation model (U-Net), a novel semantic segmentation model named VGG16-UNet-CA was established, which incorporated the U-Net, VGG16 model, Bilinear interpolation, and Coordinate attention (CA). Moreover, the high-precision binary images obtained by the established model were imported into Open Source Computer Vision Library (OpenCV) to calculate the particle shape indicators rapidly. Additionally, the cyclic loading tests were carried out to characterize the macro deterioration of graded gravel based on the mechanical indicator dynamic stiffness K, and the key factor of deterioration was explored through the established machine vision method. Finally, biaxial compression models with different deterioration degree were established by Discrete Element Method (DEM) to explore the micro deterioration mechanism. It was found that the segmentation accuracy indicators F1-score, Mean Pixel Accuracy (MPA) and Mean Intersection over Union (MIoU) of VGG16-UNet-CA were 98.56 %, 97.89 % and 98.01 %, which were higher than U-Net, SegNet, PSPNet, and DeepLabv3+. During the cyclic loading, the changes of the equivalent particle diameter De and slenderness ratio Ei of coarse particles in three filler groups were not significant, while the average values of roundness Rc increased by 28 %, 31 % and 25 %, indicating that the coarse particle abrasion was the key factor of deterioration. In the DEM simulation, with the increase of Rc, the mechanical strength of graded gravel decreased and the micro indicators (e.g., sliding rate, strong force chains, and anisotropy parameter an) inside the fillers gradually decreased, while the particle rotation continuously increased. Consequently, the interlocking ability of coarse particles decreased, leading to the bearing capacity and structural stability of particle skeleton reduced. This study not only provides a novel approach to investigate the deterioration mechanism of graded gravel under cyclic loading, which contributes to revealing the generation mechanism of high-speed railway subgrade defects in service, but also provides theoretical support for subgrade reinforcement.http://www.sciencedirect.com/science/article/pii/S2214509524011628High-speed railway subgradeGraded gravelPerformance deteriorationMachine visionCyclic loadingDEM
spellingShingle Tai-feng Li
Xian-pu Xiao
Rong-hui Yan
Kang Xie
Jia-shen Li
Ruo-han Dai
Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision
Case Studies in Construction Materials
High-speed railway subgrade
Graded gravel
Performance deterioration
Machine vision
Cyclic loading
DEM
title Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision
title_full Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision
title_fullStr Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision
title_full_unstemmed Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision
title_short Research on deterioration mechanism of graded gravel in high-speed railway subgrade layer based on machine vision
title_sort research on deterioration mechanism of graded gravel in high speed railway subgrade layer based on machine vision
topic High-speed railway subgrade
Graded gravel
Performance deterioration
Machine vision
Cyclic loading
DEM
url http://www.sciencedirect.com/science/article/pii/S2214509524011628
work_keys_str_mv AT taifengli researchondeteriorationmechanismofgradedgravelinhighspeedrailwaysubgradelayerbasedonmachinevision
AT xianpuxiao researchondeteriorationmechanismofgradedgravelinhighspeedrailwaysubgradelayerbasedonmachinevision
AT ronghuiyan researchondeteriorationmechanismofgradedgravelinhighspeedrailwaysubgradelayerbasedonmachinevision
AT kangxie researchondeteriorationmechanismofgradedgravelinhighspeedrailwaysubgradelayerbasedonmachinevision
AT jiashenli researchondeteriorationmechanismofgradedgravelinhighspeedrailwaysubgradelayerbasedonmachinevision
AT ruohandai researchondeteriorationmechanismofgradedgravelinhighspeedrailwaysubgradelayerbasedonmachinevision