轮胎带束层接头缺陷在线检测研究  

Online detection of defects in tire belt joints

作　　者：苑诗帅 焦冬梅[1] 杨世豪 王同坤 杨化林[1] YUAN ShiShuai;JIAO DongMei;YANG ShiHao;WANG TongKun;YANG HuaLin(College of Mechanical and Electrical Engineering,Qingdao University of Science and Technology,Qingdao 266000,China)

机构地区：[1]青岛科技大学机电工程学院,青岛266000

出　　处：《北京化工大学学报（自然科学版）》2022年第4期96-104,共9页Journal of Beijing University of Chemical Technology(Natural Science Edition)

基　　金：山东省自然科学基金(ZR2019MEE102)。

摘　　要：轮胎带束层在线贴合时,接头对接质量是影响轮胎安全性的关键因素,也是企业质量把控的关键环节。目前企业中对于带束层接头质量的检测方式主要是人工及少部分传统视觉检测,人工检测的评价结果受人的主观意识影响较大,导致误判率、错判率高,而传统视觉检测因为胎面粗糙度、接头缺陷性质等因素,容易出现误检率高、效率低等现象。为解决上述问题,提出一种基于机器视觉及线激光辅助的轮胎带束层接头缺陷在线检测技术,建立了一种将积分法初定位与斜率均值法精提取相结合的特征点提取方法,有效消除了带束层表面粗糙度对特征点提取精度的干扰;建立了精度校正模型,减少了伪特征点对测量精度的影响。最后通过实验验证了本文方法应用于带束层接头缺陷在线检测的有效性,实验结果表明所提方法的平均检测时间为3.35 s、误检率低于5%,相较于传统的检测方法具有更高的效率与准确率。The quality of the joint fittings is a key factor affecting the safety of tires when the tire belt is laminated online,and is also a key link in enterprise quality control.Current detection of the quality of enterprises with belt joints mainly involves manual and traditional visual methods.The evaluation results of manual detection are subjectively influenced by humans,which leads to high misjudgment and error rates,while traditional visual detection leads to a high misjudgment rate and low efficiency because of factors such as tread roughness and nature of joint defects.In light of the above problems,this paper proposes a machine vision and line-laser-assisted online method for the detection of tire belt joint defects.A feature point extraction method combining the initial localization of the integration method and the fine extraction of the slope mean method is established,which effectively eliminates interference by the surface roughness of the tire belt on the feature point extraction accuracy.An accuracy correction model is established to reduce the influence of pseudo-feature points on detection accuracy.The effectiveness of the method in the online detection of defects in the tire belt joints is verified by experiment.The experimental results show that the average detection time of the proposed method is 3.35 s,and the false detection rate is less than 5%,which is more efficient and accurate than traditional detection methods.

关 键 词：带束层 缺陷检测 机器视觉 线激光 特征点提取 

分 类 号：TN219[电子电信—物理电子学]