Fiber distribution in UHPC under different influencing factors evaluated with a novel method based on deep learning

IF 7.4 1区 工程技术 Q1 CONSTRUCTION & BUILDING TECHNOLOGY
Chen Shen , Desha Tang , Peiyi Wang , Zhaoqiu Lyu , Mingtao Zhang , Baoming Liu , Changhui Yang , Linwen Yu
{"title":"Fiber distribution in UHPC under different influencing factors evaluated with a novel method based on deep learning","authors":"Chen Shen ,&nbsp;Desha Tang ,&nbsp;Peiyi Wang ,&nbsp;Zhaoqiu Lyu ,&nbsp;Mingtao Zhang ,&nbsp;Baoming Liu ,&nbsp;Changhui Yang ,&nbsp;Linwen Yu","doi":"10.1016/j.conbuildmat.2024.139350","DOIUrl":null,"url":null,"abstract":"<div><div>The dispersion of fibers in UHPC significantly affects its mechanical properties. Currently, conventional image analysis methods are employed to evaluate fiber dispersion. However, due to the settling of steel fibers in UHPC and the subjectivity of image processing, the evaluation of fiber dispersion is neither sufficiently thorough nor accurate. In this study, removal method was used to optimize a deep learning-based U-net model for fiber identification. And sinkage coefficient was proposed as a complementary index for evaluating fiber dispersion. Subsequently, the accuracy of both the identification technique and evaluation method was separately validated. Then, the image processing techniques and evaluation method proposed in this study are applied to investigate the impact of various factors on fiber distribution in UHPC. The results indicate that the optimized deep learning-based network accomplished batch image semantic segmentation and image denoising. This approach mitigated the subjectivity during threshold selection and minimized the influence of low-quality images on fiber identification. The proposed sinkage coefficient can reflect the impact of fiber sinking on its performance. This coefficient associate with distribution and orientation coefficients can evaluate fiber dispersion comprehensively. Experimental investigations revealed that during vibration, fibers in UHPC exhibit sinking or floating phenomena. The consistency of the slurry emerges as a pivotal factor affecting fiber dispersion. Lower consistency results in pronounced fiber sinking, whereas higher consistency induce fiber floating. The incorporation of coarse aggregate facilitates the formation of a network structure with steel fibers, thereby ameliorating fiber sinking or floating tendencies, further affecting the fiber dispersion. Additionally, the increase in vibration time is also detrimental to the uniform dispersion of fibers.</div></div>","PeriodicalId":288,"journal":{"name":"Construction and Building Materials","volume":"457 ","pages":"Article 139350"},"PeriodicalIF":7.4000,"publicationDate":"2024-11-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Construction and Building Materials","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0950061824044921","RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CONSTRUCTION & BUILDING TECHNOLOGY","Score":null,"Total":0}
引用次数: 0

Abstract

The dispersion of fibers in UHPC significantly affects its mechanical properties. Currently, conventional image analysis methods are employed to evaluate fiber dispersion. However, due to the settling of steel fibers in UHPC and the subjectivity of image processing, the evaluation of fiber dispersion is neither sufficiently thorough nor accurate. In this study, removal method was used to optimize a deep learning-based U-net model for fiber identification. And sinkage coefficient was proposed as a complementary index for evaluating fiber dispersion. Subsequently, the accuracy of both the identification technique and evaluation method was separately validated. Then, the image processing techniques and evaluation method proposed in this study are applied to investigate the impact of various factors on fiber distribution in UHPC. The results indicate that the optimized deep learning-based network accomplished batch image semantic segmentation and image denoising. This approach mitigated the subjectivity during threshold selection and minimized the influence of low-quality images on fiber identification. The proposed sinkage coefficient can reflect the impact of fiber sinking on its performance. This coefficient associate with distribution and orientation coefficients can evaluate fiber dispersion comprehensively. Experimental investigations revealed that during vibration, fibers in UHPC exhibit sinking or floating phenomena. The consistency of the slurry emerges as a pivotal factor affecting fiber dispersion. Lower consistency results in pronounced fiber sinking, whereas higher consistency induce fiber floating. The incorporation of coarse aggregate facilitates the formation of a network structure with steel fibers, thereby ameliorating fiber sinking or floating tendencies, further affecting the fiber dispersion. Additionally, the increase in vibration time is also detrimental to the uniform dispersion of fibers.
基于深度学习的UHPC纤维分布评估方法
纤维在UHPC中的分散对其力学性能有显著影响。目前,光纤色散的评价主要采用传统的图像分析方法。然而,由于钢纤维在UHPC中的沉降和图像处理的主观性,对纤维弥散度的评价不够彻底和准确。在本研究中,采用去除方法对基于深度学习的U-net模型进行纤维识别优化。并提出了下沉系数作为评价纤维色散的补充指标。随后,分别验证了鉴定技术和评价方法的准确性。然后,应用本文提出的图像处理技术和评价方法,研究了各种因素对UHPC中光纤分布的影响。结果表明,优化后的深度学习网络完成了批量图像语义分割和图像去噪。该方法减轻了阈值选择过程中的主观性,最大限度地减少了低质量图像对光纤识别的影响。提出的下沉系数可以反映光纤下沉对其性能的影响。该系数与分布系数和取向系数相结合,可以综合评价光纤的色散。实验研究表明,在振动过程中,UHPC中的纤维会出现下沉或漂浮现象。浆料的稠度是影响纤维分散的关键因素。低稠度导致纤维下沉,高稠度导致纤维漂浮。粗骨料的掺入有利于与钢纤维形成网状结构,从而改善纤维的下沉或漂浮趋势,进而影响纤维的分散性。此外,振动时间的增加也不利于纤维的均匀分散。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Construction and Building Materials
Construction and Building Materials 工程技术-材料科学:综合
CiteScore
13.80
自引率
21.60%
发文量
3632
审稿时长
82 days
期刊介绍: Construction and Building Materials offers an international platform for sharing innovative and original research and development in the realm of construction and building materials, along with their practical applications in new projects and repair practices. The journal publishes a diverse array of pioneering research and application papers, detailing laboratory investigations and, to a limited extent, numerical analyses or reports on full-scale projects. Multi-part papers are discouraged. Additionally, Construction and Building Materials features comprehensive case studies and insightful review articles that contribute to new insights in the field. Our focus is on papers related to construction materials, excluding those on structural engineering, geotechnics, and unbound highway layers. Covered materials and technologies encompass cement, concrete reinforcement, bricks and mortars, additives, corrosion technology, ceramics, timber, steel, polymers, glass fibers, recycled materials, bamboo, rammed earth, non-conventional building materials, bituminous materials, and applications in railway materials.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信