|本期目录/Table of Contents|

[1]沈浩,江臣,陈宇文,等.基于深度学习的钢桁架桥螺栓病害智能识别方法[J].南京工业大学学报(自然科学版),2020,42(05):608-615.[doi:10.3969/j.issn.1671-7627.2020.05.009]
 SHEN Hao,JIANG Chen,CHEN Yuwen,et al.Intelligent identification method for bolt diseases of steel truss bridge based on deep learning model[J].Journal of NANJING TECH UNIVERSITY(NATURAL SCIENCE EDITION),2020,42(05):608-615.[doi:10.3969/j.issn.1671-7627.2020.05.009]
点击复制

基于深度学习的钢桁架桥螺栓病害智能识别方法()
分享到:

《南京工业大学学报(自然科学版)》[ISSN:1671-7627/CN:32-1670/N]

卷:
42
期数:
2020年05期
页码:
608-615
栏目:
出版日期:
2020-09-20

文章信息/Info

Title:
Intelligent identification method for bolt diseases of steel truss bridge based on deep learning model
文章编号:
1671-7627(2020)05-0608-08
作者:
沈浩1江臣2陈宇文3王国香3李枝军3
1.苏交科集团股份有限公司,江苏 南京 210017; 2.江苏省交通工程建设局,江苏 南京 210004; 3.南京工业大学 土木工程学院,江苏 南京 211800
Author(s):
SHEN Hao1 JIANG Chen2 CHEN Yuwen3 WANG Guoxiang3 LI Zhijun3
1. Jiangsu Transportation Research Institute Co. Ltd., Nanjing 210017, China; 2. Jiangsu Provincial Transportation Engineering Construction Bureau, Nanjing 210004, China; 3. College of Civil Engineering,Nanjing Tech University, Nanjing 211800, China
关键词:
钢桁架桥 螺栓病害 无人机航拍 图像处理 迁移学习 神经网络
Keywords:
steel truss bridges bolt diseases unmanned aerial vehicle(UAV)video image processing transfer learning neural networks
分类号:
U443.32
DOI:
10.3969/j.issn.1671-7627.2020.05.009
文献标志码:
A
摘要:
为了提高钢桁架桥螺栓病害检测和识别效率、完善分析方法,本文在无人机航拍视频的基础上,提出了一种基于深度学习的螺栓病害智能识别方法。通过混合高斯算法、Canny边缘检测、最小包围圆算法等传统图像处理手段对航拍视频进行预处理,实现钢桁架桥螺栓图像的批量化提取,并通过对螺栓图像采取缩放、旋转、变形等措施拓展螺栓图像的样本数; 采用迁移学习引入深度学习模型INCEPTION-V3,经过训练,当螺栓数据测试集上的准确率大于95%时,可满足工程精度需求; 并将该方法应用于实际工程,当把0.8设置为计算螺栓病害概率的分割点时,该方法对螺栓病害具有较好的识别效果,同时能够实现自动化、智能化,避免人为主观判断带来的影响。
Abstract:
To improve the efficiency of detection and identification of bolt damage of steel truss bridges and optimize the analysis methods, a method for intelligent identification of bolt disease based on deep learning was proposed, using unmanned aerial vehicle(UAV)video. The UAV video was pre-processed by traditional image processing methods such as Gaussian algorithm, Canny edge detection and minimum bounding circle algorithm. Batch extraction of steel truss bolt images was realized, and measures such as scaling, rotation and deformation of bolt images were taken to expand the number of samples of the bolt image. The deep learning model of INCEPTION-V3 was introduced to transfer learning. After training, the accuracy rate on the bolt data test set was greater than 95%, which could meet the needs of engineering accuracy. Results showed that when 0.8 was set as the dividing point for calculating the probability of disease, the method had a better recognition effect on bolt disease, and could realize automation and intelligence, and avoid the impact of human subjective judgment.

参考文献/References:

[1] 王子斌,郭勤涛,展铭.基于模型的螺栓松动状态监测方法[J].振动.测试与诊断,2019,39(2):281.
[2] 付宜东,陈烨,毛泽亮,等.高强度螺栓不同拧入深度对螺栓球节点受拉承载力的影响[J].南京工业大学学报(自然科学版),2017,39(1):106.
[3] 黄炳生,付宜东,陈烨,等.螺栓球节点高强度螺栓拧入深度试验研究[J].建筑钢结构进展,2016,18(4):28.
[4] 陈国达,郗枫飞,计时鸣,等.螺栓无损检测方法综述[J].制造技术与机床,2017(11):22.
[5] 钟继卫,王波,王翔,等.桥梁智能检测技术研究与应用[J].桥梁建设,2019,49(增刊1):1.
[6] 张军.基于数字图像处理的瓷砖表面缺陷检测研究[D].淄博:山东理工大学,2018.
[7] 冯敏,罗旺,余磊,等.适用于无人机巡检图像的输电线路螺栓检测方法[J].电力科学与技术学报,2018,33(4):135.
[8] 詹琦梁,陈胜勇,胡海根,等.一种结合多种图像分割算法的实例分割方案[J].小型微型计算机系统,2020,41(4):837.
[9] 杜玉龙,李建增,张岩,等.基于深度交叉CNN和免交互GrabCut的显著性检测[J].计算机工程与应用,2017,53(3):32.
[10] 张晗,钱育蓉,王跃飞,等.基于图像边缘增强与弱化的边缘检测[J].计算机工程与设计,2019,40(11):3106.
[11] 张英明,李永亮,张翼鹏,等.激光光斑边缘检测算子的研究进展[J].量子光学学报,2019,25(1):109.
[12] 张淑芳,刘孟娅,韩泽欣,等.基于Retinex增强的单幅LDR图像生成HDR图像方法[J].计算机辅助设计与图形学学报,2018,30(6):1015.
[13] 张光华,王福豹,段渭军.基于GAN的天文图像生成研究[J].西北工业大学学报,2019,37(2):315.
[14] 王小鹏,渠燕红,马鹏,等.基于头部图像特征的人流计数方法[J].铁道学报,2019,41(2):78.
[15] 周飞燕,金林鹏,董军.卷积神经网络研究综述[J].计算机学报,2017,40(6):1229.
[16] 杨真真,匡楠,范露,等.基于卷积神经网络的图像分类算法综述[J].信号处理,2018,34(12):1474.
[17] 马杰.基于卷积神经网络的无人机侦察图像识别[D].北京:北京邮电大学,2019.
[18] 李楠,蔡坚勇,李科,等.基于多Inception结构的卷积神经网络人脸识别算法[J].计算机系统应用,2020,29(2):157.
[19] 吴迪,侯凌燕,刘秀磊,等.一种改进的深度神经网络的花卉图像分类[J].河南大学学报(自然科学版),2019,49(2):192.
[20] 樊彦国,柴江龙,许明明,等.基于ORB与RANSAC融合改进的图像配准[J].光学精密工程,2019,27(3):702.

备注/Memo

备注/Memo:
收稿日期:2020-02-27
基金项目:江苏省交通运输科技项目(2017-2-10,2019Y18)
作者简介:沈浩(1987—),男,工程师,E-mail:215397349@qq.com; 李枝军(联系人),副教授,E-mail:lizhijun@njtech.edu.cn.
引用格式:沈浩,江臣,陈宇文,等.基于深度学习的钢桁架桥螺栓病害识别方法[J].南京工业大学学报(自然科学版),2020,42(5):608-615.
SHEN Hao, JIANG Chen, CHEN Yuwen, et al. Intelligent identification method for bolt diseases of steel truss bridge based on deep learning model[J].Journal of Nanjing Tech University(Natural Science Edition),2020,42(5):608-615..
更新日期/Last Update: 2020-09-24