基于计算机视觉技术的螺栓松动损伤识别研究  

Loosenedbolt damage detection based on computer vision

作　　者：吴杰 程帆 黄楚越 WU Jie;CHENG Fan;HUANG Chuyue(School of Civil Engineering and Architecture,Wuhan Polytechnic University,Wuhan 430023,China;Hubei Assembly Building Engineering Technology Research Center,Wuhan 430023,China)

机构地区：[1]武汉轻工大学土木工程与建筑学院,武汉430023 [2]湖北省装配式建筑工程技术研究中心,武汉430023

出　　处：《武汉轻工大学学报》2023年第3期91-98,共8页Journal of Wuhan Polytechnic University

基　　金：湖北省住房和城市建设厅项目(2020年城乡建设发展-01);武汉轻工大学校立科研项目(2021Y47)。

摘　　要：螺栓连接是钢结构的主要连接方式。钢结构在服役过程中,因长期循环荷载作用等外界环境影响,螺栓连接将不可避免发生松动、损伤等病害。如何快速有效地识别螺栓松动等损伤对于确保结构的完备以及安全至关重要。基于计算机视觉技术,以特征追踪算法为基础,通过自动识别不同时间采集到的相片中的特征点的分布情况识别结构中松动的螺栓,建立了钢结构螺栓连接状态的非接触无损检测方法。首先找出相片中的特征点,然后追踪这些特征点在不同时间下的状态,绘出特征点的运动轨迹,根据运动轨迹的状态判断与之相关联的螺栓是否有损,如若螺栓松动,则与该螺栓关联的特征点会显示出唯一的运动状态,通过标记这些有明显运动的特征点则可以有效定位出有损的螺栓。以一钢结构移梁装置的螺栓群为试验对象,通过扭转6个螺栓中的1个,采用建议方法有效地识别了存在松动的螺栓。Bolted steel joints are one of the most common connections in steel structures.Under long-term cyclic loadings,the bolt steel joints of the steel structures have many damages and monitoring the state of the bolt steel joints is particularly important to ensure the functionality and safety of the structures.Among all the factors of bolt steel joints damage,bolt loosening is reported to be the most important one.Therefore,for structural health assessment,it is critically important to timely identify bolt looseness.In this paper,a monitoring approach for bolt loosening dectection based on computer vision with non-contact non-damage is proposed.Based on feature tracking algorithm,the loosened bolts are automatically identified through the distribution of features points from multiple images.The method relies on the features of the images,then track the status of them,obtain the trajectories of feature points.According to the movement of the feature points,the loosened bolt can be identified.If the bolt is loosened,feature points associated with the loosened bolt would exhibit a unique movement pattern,and by marking the movement pattern,the damaged bolt can be located.Finally,taking one bolt group of a beam moving device as an object,the loosened bolt is detected successfully with the proposed algorithm.

关 键 词：计算机视觉 结构健康监测 螺栓松动识别 特征追踪 Shi-Tomasi特征点 Kanade-Lucas-Tomasi特征追踪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]