摘要
在生产过程中,预制构件尺寸不合格问题将导致其在施工现场无法顺利安装,从而影响工期。为推进预制构件智能化生产的进程,以预制叠合板为例,基于卷积神经网络研究生产过程中的智能检测方法,在生产流水线上设计并安装图像采集系统,建立预制叠合板尺寸检测数据集。通过YOLOv5算法实现对混凝土底板、预埋PVC线盒及外伸钢筋的识别,并以固定磁盒作为基准参照物进行尺寸检测误差分析,实现混凝土底板尺寸、预埋PVC线盒坐标的检测,在降低训练数据集参数规模的工况下保持较高的识别精度。结果表明:该方法可以有效检测预制叠合板的底板数量和尺寸、预埋PVC线盒数量和坐标,并实现弯折方向不合格的外伸钢筋检测,并能降低人工成本,提高检测精度,加快检测速度,提高预制叠合板的出厂质量。
The unqualified size of prefabricated component in the production process will lead to the failure of the installation on the construction site,and affect the construction period.In order to promote the process of intelligent production of prefabricated components.Based on a convolutional neural network,the prefabricated laminated board is used as an example to study the intelligent detection method of the production process.Design and install an image acquisition system on the production line,establish a prefabricated laminated board detection data set,and use the YOLOv5 algorithm to detect the concrete plate,the embedded PVC junction box and the overhanging steel bar.The fixed magnetic box is used as the benchmark to analyze the detection error of the dimension of the concrete plate and the coordinate of the embedded PVC junction box,and maintains a high recognition accuracy with a smaller parameter scale of the training data set.The result shows that the method can effectively detect the number and dimension of the concrete plate,the number and coordinate of the embedded PVC junction box,and detect the overhanging steel bar of unqualified bending direction.The method can reduce labor costs,improve detection accuracy,speed up detection process,and improve the delivery quality of prefabricated laminated board.
作者
姚刚
廖港
杨阳
李青泽
魏伏佳
YAO Gang;LIAO Gang;YANG Yang;LI Qingze;WEI Fujia(Key Laboratory of New Technology for Construction of Cities in Mountain Area,Ministry of Education,School of Civil Engineering,Chongqing University,Chongqing 400045,P.R.China;China Machinery China United Engineering Co.,Ltd.,Chongqing 400050,P.R.China)
出处
《土木与环境工程学报（中英文）》
CSCD
北大核心
2024年第1期93-101,共9页
Journal of Civil and Environmental Engineering
基金
国家重点研发计划(2019YFD1101005)。
关键词
预制叠合板
多目标检测
卷积神经网络
预制构件
智能化生产
prefabricated laminated board
multi-target detection
convolutional neural network
prefabricated component
intelligent production
作者简介
姚刚(1963-),男,教授,博士生导师,主要从事土木工程施工研究,E-mail:yaocqu@vip.sina.com;通信作者:杨阳,女,博士,E-mail:yy20052710@163.com。
