The use of artificial intelligence to detect defects in building structures

Author:

Knyazeva Natal'ya1,Nazojkin Evgenij2,Orekhov Aleksej3

Affiliation:

1. Moscow State University of Civil Engineering (National Research University)

2. Russian Biotechnological University (ROSBIOTECH)

3. Rossiyskiy biotehnologicheskiy universitet (ROSBIOTEH)

Abstract

Monitoring the technical condition of structures is the most important task aimed at improving the reliability and safety of buildings and structures. During the survey, a set of tasks arises to assess visible defects and damages, the solution of which requires the experience and attention of structural survey specialists. Often the omission of visible defects is the most common mistake when examining the engineering and technical condition of a building. Technical vision, as a method of classifying objects in images, can significantly improve the efficiency of visual inspection and reduce the number of errors on the object. In this paper, an algorithm for detecting damage to reinforced concrete structures based on a convolutional neural network model created in the Python programming language is investigated. The neural network was trained and tested on real defects of a monolithic reinforced concrete building. According to the results of the work, the high efficiency of artificial intelligence in determining defects and damages in the framework of the survey of the engineering and technical condition of monolithic reinforced concrete structures of a building under construction was revealed. Automation of works on visual inspection of building structures is a promising direction for the development of artificial intelligence.

Publisher

RIOR Publishing Center

Subject

Industrial and Manufacturing Engineering,Polymers and Plastics,Business and International Management

Reference13 articles.

1. Shi, С. and Pun, С.-М., “Supcrpixcl-bascd 3D deep neural networks for hypcrspcctral image classification”, Pattern Rccognit., Vol. 74, Pages 600-616, 2018., Shi, C. and Pun, C.-M., “Supcrpixcl-bascd 3D deep neural networks for hypcrspcctral image classification”, Pattern Rccognit., Vol. 74, Pages 600-616, 2018.

2. Li, J., Xia, C. and Chen, X., “A Benchmark Dataset and Salicncy-Guidcd Stacked Autocncodcrs for Video-Based Salient Object Detection”, IEEE Trans. Image Process., Vol. 27, Issue 1, Pages 349- 364, 2018., Li, J., Xia, C. and Chen, X., “A Benchmark Dataset and Salicncy-Guidcd Stacked Autocncodcrs for Video-Based Salient Object Detection”, IEEE Trans. Image Process., Vol. 27, Issue 1, Pages 349- 364, 2018.

3. Наумов А.Е. Совершенствование технологии строительства и технической экспертизы с использованием аппаратно-программного комплекса автоматизированной дефектоскопии / А.Е. Наумов, Д.А. Юдин, А.В. Долженко // Вестник Белгородского государственного технологического университета. ВГ Шухов. – 2019. – № 4. – С. 61-69., Naumov A.E. Improvement of construction technology and technical expertise using hardware and software complex of automated flaw detection / A.E. Naumov, D.A. YUdin, A.V. Dolzhenko // Bulletin of BSTU named after V.G. Shukhov. – 2019. – № 4. – S. 61-69.

4. Сикорский О.С. Обзор сверточных нейронных сетей для задачи классификации изображений / О.С. Сикорский // Новые информационные технологии в автоматизированных системах. – 2017. – № 3. – С. 15-23., Sikorskij O.S. Overview of convolutional neural networks for image classification problem / O.S. Sikorskij // New information technologies in automated systems. – 2017. – № 3. – S. 15-23.

5. Соснин А.С., Суслова И.А. Функции активации нейронной сети: сигмовидная, линейная, ступенчатая, ReLu, tan // Наука. Информатизация. Технологии. Образование. - Екатеринбург:, 2019. - С. 237-246., Sosnin A.S., Suslova I.A. Neural network activation functions: sigmoid, linear, stepwise, ReLu, tan // The science. Informatization. Technologies. Education. - Ekaterinburg:, 2019. - S. 237-246.

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3