The position of prestressed ducts has a significant impact on the bearing capacity of prestressed concrete members. A position detection method for ducts based on 3D laser scanning and deep learning is proposed to improve the efficiency and accuracy of ducts position detection. Firstly, a dataset containing prestressed ducts is produced by the 3D laser scanning method. Then, the ducts point cloud is obtained by an efficient large-scale 3D point cloud semantic segmentation network (RandLA-Net). The centroid of the ducts point cloud slice is obtained by using the bracket box midpoint method and circle fitting constraints method. Finally, the centerline of the ducts is fitted by the back-propagation neural network (BP network). The detection results of actual prestressed concrete box girder duct position using this method show that even under the interference of complex surrounding environments such as box girder reinforcing steel skeleton and construction tire frame, this method can obtain the complete prestressed duct position. The maximum detection error is less than ±10 mm, which meets the requirements of relevant construction specifications.
本文的实验环境为64位Ubuntu18.04操作系统,CPU为Intel(R) i7-7800X,内存大小为16 G,显卡为NVIDIA GeForce RTX 3090,显存大小为24 G. 深度学习使用CUDA11.4加速GPU计算,虚拟环境为基于python3.6的Tensorflow2.6-pu. Batchsize设置为6,epochs为100,初始学习率为0.01,学习率每个epoch衰减5%,模型单次输入点数为40 960,最近邻点K值为16.
2.1.2 评价指标
点云的语义分割性能常采用总体准确率(overall accuracy,OA)、平均准确率(mean accuracy,mAcc)、交并比(intersection over union,IoU)以及平均交并比(mean intersection over union,mIoU)作为评估指标. 代表预测正确的点云数量占全部点云数量的比例,为各个类别中正确分割的点云个数占实际点云数的比例总和的平均值,是每一类点云在其实际和分割结果中预测正确的准确度,而表征了所有类别交并比之和的平均值,从全局上衡量了网络分割的准确性. 指标具体的计算公式如下:
GUOY .The construction technology of PC-beam pre-stressed duct’s three-dimensional integrate precision location[J].Railway Construction Technology,2013(1):55-58.(in Chinese)
[3]
高建军 .基于有效预应力检测的预应力混凝土桥梁评价方法[D].西安:长安大学,2007.
[4]
GAOJ J .Evaluation method for prestressed concrete bridge based on detected effective prestressed[D].Xi’an:Chang’an University,2007.(in Chinese)
[5]
成琛 .大跨径PC桥梁弯曲孔道有效预应力理论分析与试验研究[D].武汉:武汉理工大学,2011.
[6]
CHENGC .Theoretical analysis and experimental study on effective prestress of the long span PC bridge in curved duct[D].Wuhan:Wuhan University of Technology,2011.(in Chinese)
CUIX, CHENY, WUT,et al .Research on the position probe of prestressed pipe based on the ground penetrating radar method[J].Shanghai Highways,2016(2):61-64.(in Chinese)
CHENH T, LINC .Research on location and data processing of ground penetrating radar for prestressed pipe in bridges[J].Chinese Journal of Engineering Geophysics,2020,17(1):101-106.(in Chinese)
PANH J, HUANGF W .Application of GPR technology in the location detection of bridge prestressed pipeline[J].Journal of East China Jiaotong University,2012,29(1):67-70.(in Chinese)
CHENW, QUG Z, LIW X .Application of impact echo method in positioning of prestressed pipeline[J].Highway Traffic Science and Technology (Applied Technology Edition),2018(10):55-56.(in Chinese)
BANP, WANGJ, ZHANGT B, et al. Comprehensive benefit evaluation of positioning device for advanced prestressed tendons based on AHP[C]//Proceedings of the 2020 National Civil Engineering Construction Technology Exchange Conference (Part Ⅱ). Beijing, 2020: 190-193. (in Chinese)
[17]
KIMM K, SOHNH, CHANGC C .Automated dimensional quality assessment of precast concrete panels using terrestrial laser scanning[J].Automation in Construction,2014,45:163-177.
[18]
WANGQ, CHENGJ C P, SOHNH. Automated estimation of reinforced precast concrete rebar positions using colored laser scan data[J]. Computer-Aided Civil and Infrastructure Engineering, 2017, 32(9): 787-802.
WUW Q, WANGX Y, LIUH Y,et al .Research on appearance and dimension detection method based on laser point clouds for precast concrete box girder[J].Bridge Construction,2023,53(4):25-32.(in Chinese)
[21]
邓奕光 .基于深度学习的三维激光点云轨道对象语义分割与提取[D].武汉:武汉大学,2019.
[22]
DENGY G .Semantic segmentation and extraction of 3D laser point cloud orbit objects based on deep learning[D].Wuhan:Wuhan University, 2019.(in Chinese)
[23]
ZOUY L, WEINACKERH, KOCHB .Towards urban scene semantic segmentation with deep learning from LiDAR point clouds:a case study in Baden-Württemberg,Germany[J].Remote Sensing, 2021, 13(16): 3220.
[24]
HUQ Y, YANGB, XIEL H,et al. Learning semantic segmentation of large-scale point clouds with random sampling[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(11): 8338-8354.
BAOJ, ZHAOJ Y, ZHOUH Y. Study on method of curve simulation based on BP network[J]. Computer Engineering and Design, 2005, 26(7): 1840-1841.(in Chinese)
WUW G, TIANS Y, ZHANGZ Y, et al. Research on surface geometry parameter recognition and model reconstruction of uneven road[J].Automotive Engineering,2023,45(2):273-284.(in Chinese)
基金资助
国家自然科学基金资助项目(51978392)
National Natural ScienceFoundation of China(51978392)
国家自然科学基金青年科学基金项目(52208317)
Young Scientists Fund of the National Natural Science Foundation of China(52208317)
上海市科技攻关基金项目(20dz1202100)
Social Development Science and Technology Research Project of Shanghai(20dz1202100)