直缝焊管焊缝噘嘴检测与识别技术研究

发布时间：2018-05-21 17:33

本文选题：直缝焊管 + 焊缝噘嘴　；参考：《河南理工大学》2016年硕士论文

【摘要】：本文主要针对直缝焊管焊缝噘嘴的检测技术做了深入的研究。首先阐述了直缝焊管焊缝噘嘴的成因,然后介绍了国内外研究现状,并和现有的检测方法做比较,确定了采用激光扫描技术来对直缝焊管焊缝噘嘴缺陷进行检测的方法。该方法是利用激光扫描仪对直缝焊管表面的三维轮廓数据进行采集,然后对数据进行一系列的算法处理,最后选用随机Hough变换检测圆的方法提取焊管缺陷的特征,以此来判断直缝焊管的焊缝是否存在噘嘴缺陷。为了提高随机Hough变换的算法效率以及其准确性,需要对焊管表面三维轮廓数据中的圆弧数据和焊缝处特征数据进行分离,利用干净的圆弧数据来进行随机Hough变换提取缺陷特征,这时提出了利用K均值聚类算法来对随机Hough变换算法来进行优化。通过实验分析可知,由于焊管表面三维轮廓数据的特点,直接对这点数据点进行K均值聚类的效果并不理想。为了提高聚类的准确性,提出了利用差分曲率算法来优化K均值聚类算法的方法。通过差分曲率算法计算出所有离散点的曲率,通过对这些曲率的分布进行研究,发现这些曲率值的分布有一定的规律,焊缝处数据点的曲率和焊缝两侧圆弧数据的曲率值的大小有明显的区别,利用这一特点,可以通过K均值聚类算法将它们划分开,然后在对圆弧数据进行随机Hough变换检测圆。通过实验分析,经过优化之后的K均值聚类算法以及随机Hough变换算法的准确性和实时性都有明显提高。最后,基于本文中提出的直缝焊管焊缝噘嘴检测方法,在MATLAB/GUI环境下设计了一个直缝焊管焊缝噘嘴检测系统,实验结果表明该系统能够有效的对直缝焊管焊缝噘嘴进行判断。
[Abstract]:In this paper, the detection technology of welded seam pout of straight welded pipe is studied in detail. In this paper, the causes of weld pout of straight welded pipe are introduced, then the research status at home and abroad is introduced, and compared with the existing inspection methods, the method of detecting the weld pout defects of straight welded pipe by laser scanning technology is determined. This method uses laser scanner to collect 3D profile data of straight welded pipe surface, then carries on a series of arithmetic processing to the data, finally selects the random Hough transformation to detect the circle method to extract the characteristic of the welded pipe defect. In order to determine whether the weld seam of straight welded pipe has pouting defects. In order to improve the algorithm efficiency and accuracy of random Hough transform, it is necessary to separate arc data from weld seam feature data from 3D profile data of welded pipe surface, and to extract defect feature by random Hough transform using clean arc data. A K-means clustering algorithm is proposed to optimize the stochastic Hough transform algorithm. According to the experimental analysis, due to the characteristics of 3D profile data of welded pipe surface, the effect of K-means clustering directly on this point data point is not ideal. In order to improve the accuracy of clustering, the differential curvature algorithm is proposed to optimize K-means clustering algorithm. The curvature of all discrete points is calculated by the difference curvature algorithm. By studying the distribution of these curvature, we find that the distribution of these curvature values has a certain rule. There are obvious differences between the curvature of the data points at the weld and the curvature values of the arc data on both sides of the weld. By using this feature, they can be divided by K-means clustering algorithm, and then the circle is detected by random Hough transform of the arc data. Through experimental analysis, the accuracy and real-time performance of the optimized K-means clustering algorithm and stochastic Hough transform algorithm are improved obviously. Finally, based on the method proposed in this paper, a weld pouting detection system for straight welded pipe is designed under MATLAB/GUI environment. The experimental results show that the system can effectively judge the weld pouting of straight welded pipe.
【学位授予单位】：河南理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TG441.7;TP311.13

【相似文献】