【作者】 肖勇；

【导师】 张华；

【作者基本信息】 南昌大学 ， 信息与通信工程， 2017， 硕士

【摘要】 针对当前埋弧焊基于点激光或单线激光进行焊缝跟踪存在的不足,点激光焊缝跟踪自动化程度低,单线激光焊缝跟踪存在前置误差以及高度偏差识别精度低等。提出了基于双线激光视觉传感的埋弧焊焊缝跟踪解决方案。本课题就埋弧焊基于双线激光焊缝跟踪相关技术开展研究。首先,基于现有的埋弧自动焊小车,对其进行了结构上与控制上的改进。结构上,设计了焊枪机头水平和竖直方向的传动系统,使焊枪在水平和竖直方向的调节由手动转为了自动。控制上,由主板PCM-3362、运动控制卡、图像采集卡等搭建了埋弧焊焊缝自动跟踪硬件控制系统,同时基于VC++6.0开发平台开发了埋弧焊焊缝自动跟踪软件控制系统。其次,设计了一款小巧、紧凑的双线激光视觉传感器。根据设计要求确定了传感器壳体以及内部的CCD与激光器,并建立了双线激光视觉传感器斜线焊缝偏差、倾斜焊缝偏差以及空间焊缝偏差提取数学模型,同时研究了斜线焊缝、倾斜焊缝以及空间焊缝偏差提取算法。再次,基于VC++6.0对双线激光所成的双条纹焊缝图像进行了处理。焊缝图像处理方法有平滑滤波对图像增强,KSW熵算法确定二值化阈值,Kirsch算子边缘提取,逐列搜索法中心取样以及斜率分析法特征点检测。对于图像处理得到的焊缝偏差采用了模糊控制方法。结果表明,处理一帧焊缝图像时间较短,满足焊缝跟踪实时性要求。最后,焊缝跟踪实验。对采用的图像处理算法进行了精度分析,同时进行了焊缝自动跟踪模拟试验,最后做了斜线焊缝、倾斜焊缝以及空间焊缝的焊接实验。实验结果表明,图像处理精度较高,模拟跟踪试验效果较好,实时跟踪焊接质量良好。更多还原

【Abstract】 In view of the shortcomings of current submerged arc welding based on spot laser or single line laser for seam tracking,That is to say,point laser weld tracking automation is low,While the single-line laser weld tracking there is a pre-error and a high degree of deviation recognition accuracy is low.A scheme is proposed that two-wire laser vision sensor for submerged arc welding seam tracking.In this paper,having carried out research on submerged arc welding based on two-wire laser weld tracking technologyFirst of all,based on the existing submerged arc welding car,have improve its structural and control.In structure.Designed the torch head horizontal and vertical direction of the transmission system,So that the welding gun in the horizontal and vertical direction of the adjustment from the manual to automatic.In Control,By the motherboard PCM-3362,motion control card,image acquisition card to build a submerged arc welding seam automatic tracking hardware control system.At the same time,the automatic tracking software control system of submerged arc welding seam was developed based on VC++6.0 development platform.Secondly,have designed a small,compact two-wire laser vision sensor.According to the design requirements to determine the sensor shell and the internal CCD and laser,And establishes the two-line laser vision sensor lateral deviation,height deviation and spatial deviation extraction mathematical model,At the same time,have studied the transverse deviation,height deviation and spatial deviation extraction algorithm.Once again,have processed double-line laser vision sensor made of double stripe weld image based on VC++6.0.The image processing method has a smoothing filter combined with a two-dimensional fractional-order Kalman filter for image enhancement.KSW entropy algorithm to determine the binarization threshold,Krisch operator edge extraction,Column search method to center sampling and slope analysis to feature point detection.The Fuzzy control method is adopted for the weld deviation obtained by image processing.The results show,Processing a frame time is about 20 ms,Meet the weld tracking real-time requirements.Finally,Weld tracking experiment.Have analyzed the precision of the image processing algorithm at the same time,the automatic tracking of the weld was carried out,Finally,a welding test was conducted with lateral deviations,height deviations,and elevation of the elevation.The results showed that the image processing accuracy is high,tracking test results better and welding quality is good..更多还原