具有复杂边界条件的拉索索力机器视觉识别方法  

Machine vision method for cable force identification in complex boundary conditions

作　　者：石海健 王佐才[1,2] 王令侠 辛宇[1] 段大猷 SHI Haijian;WANG Zuocai;WANG Lingxia;XIN Yu;DUAN Dayou(College of Civil Engineering,Hefei University of Technology,Hefei 230009,China;Anhui Province Road and Bridge Inspection Engineering Research Center,Hefei 230009,China;China Railway Major Bridge Engineering Group Co.,Ltd.,Wuhan 430050,China;School of Urban Construction and Transportation,Hefei University,Hefei 230601,China)

机构地区：[1]合肥工业大学土木与水利工程学院,安徽合肥230009 [2]安徽省道路与桥梁检测工程研究中心,安徽合肥230009 [3]中铁大桥局集团有限公司,湖北武汉430050 [4]合肥大学城市建设与交通学院,安徽合肥230601

出　　处：《振动工程学报》2025年第4期739-749,共11页Journal of Vibration Engineering

基　　金：国家自然科学基金资助项目(52278301,52308310)。

摘　　要：为准确识别具有复杂边界条件的拉索索力,提出了基于机器视觉和广义回归神经网络(generalized regression neural net⁃work,GRNN)的索力识别方法。采用基于相位的运动放大算法和亚像素边缘定位等机器视觉技术,通过拉索振动视频提取振动位移时程并识别频率,实现拉索振动变形的多点非接触同步测量;利用有限差分法生成样本数据集,通过麻雀搜索算法(sparrow search algorithm,SSA)寻找GRNN最优光滑因子,构建SSA⁃GRNN索力识别模型以建立复杂边界条件下频率与索力的对应关系,将获取的频率信息输入模型中进行索力识别。以单根拉索为例,开展了复杂边界下拉索的数值模拟和人工激励状况下的拉索试验。结果表明,基于机器视觉和GRNN的索力识别方法可以通过振动视频准确识别频率,提高了对具有复杂边界条件的拉索索力的识别精度。In order to accurately identify cable force in complex boundary conditions,a new method of cable force identification using machine vision and generalized regression neural network(GRNN)is proposed.Machine vision technologies,such as the phase-based motion amplification algorithm and sub-pixel edge detection algorithm,are used to extract the vibration displacement time history data and identify the frequency through the cable vibration video to realize multi-point non-contact synchronous measurement of cable vibration deformation.A sample dataset is generated using the finite difference method.The smoothing factor of GRNN is obtained by the sparrow search algorithm(SSA),and a SSA-GRNN cable force prediction model is constructed,establishing the correspondence between frequencies and cable force under complex boundary conditions.The obtained frequency information is input into the model for cable force recognition.Taking a single cable as an example,the numerical simulation of the cable in complex boundary conditions and the cable test under artificial excitation condition are carried out.The results show that the cable force identification using machine vision and GRNN can accurately identify frequencies through vibration video,and improve the recognition accuracy of the cable force in complex boundary conditions.

关 键 词：索力识别 振动频率法 复杂边界条件 机器视觉 广义回归神经网络 

分 类 号：U443.38[建筑科学—桥梁与隧道工程] P234[交通运输工程—道路与铁道工程]