焊接缺陷磁光成像特征分析与探伤方法研究

发布时间：2018-04-27 22:09

本文选题：焊接缺陷 + 磁光成像　；参考：《广东工业大学》2016年硕士论文

【摘要】：焊接技术是工业制造的基础。在焊接过程中,由于受焊接现场各种干扰因素的影响,如激光功率、焊接速度、离焦量、氩气流量、高温高压以及工件表面状况等,焊接过程会出现不稳定状态,进而产生焊接缺陷,直接影响焊接质量。为保证连接材料的安全可靠且不影响被检测对象的试验性能,需采用相应的无损检测技术通过不损伤被检查材料的方式,对焊接材料的状态进行无损的检测,及时检测缺陷,杜绝潜在的危害。但常规的无损检测方法存在着一定的限制和不足,因此,探索一种基于磁光检测的新方法,用于焊接缺陷的无损检测。论文采用磁光成像无损检测焊接工件缺陷方法,结合法拉第电磁感应原理,法拉第磁致旋光效应,布儒斯特定律以及计算机机器视觉,研究偏振光发生光源,偏振光光路设计和磁光激励方式的控制参数。利用励磁场磁化焊接工件,采用检偏器检测因磁场分量在焊接缺陷位置发生变化产生的偏振光旋转变化,得到明暗变化的磁光图像。在实际焊接缺陷检测过程中,由于受到磁场强度变化和各种其他噪声的影响,采集到的磁光图像仍然会存在退化问题,清晰度较差,识别度降低,难于对焊接工件缺陷磁光图像进行透彻的分析。因此根据磁光感应成像的检测环境特点,以工件焊接缺陷为检测对象,研究焊接缺陷磁光图像的恢复算法。试验通过分析焊接工件缺陷磁光图像的特征,利用模糊集合论原理,采用改进的连续模糊增强算法提高区分度,以高强钢为检测对象,研究其表面裂纹磁光成像不均、裂纹和熔融区区分度低的不足。利用自适应快速边缘检测算法提取焊接裂纹图像的纹理特征。提取裂纹磁光图像边缘特征,提高焊接裂纹检测跟踪过程的准确性。论文采用图像处理算法对焊缝磁光图像进行特征识别并估计最优尺度,根据分类算法对提取的焊接缺陷特征进行分析和训练,构建焊接缺陷特征量并对焊接材料表面缺陷磁光图像进行自动识别。试验结果表明,运用磁光成像方法可以获取焊接材料焊接缺陷特征,并通过图像分形维数分析可识别焊缝缺陷的位置、形状和类别,同时探索了通过主成分分析方法提取焊接缺陷特征以及利用支持向量机进行焊接材料的缺陷磁光成像识别对提高分类精确度的有效性。
[Abstract]:Welding technology is the foundation of industrial manufacture. In the welding process, because of the influence of various interference factors, such as laser power, welding speed, defocusing amount, argon flow rate, high temperature and high pressure, and the surface condition of the workpiece, the welding process will appear unstable state. Then the welding defects are produced, which directly affect the welding quality. In order to ensure the safety and reliability of the connection material and not affect the test performance of the tested object, it is necessary to adopt the corresponding nondestructive testing technology to detect the defects of the welding material without damaging the inspected material. Put an end to potential hazards. However, there are some limitations and shortcomings in conventional nondestructive testing methods. Therefore, a new method based on magneto-optic testing is explored for nondestructive testing of welding defects. In this paper, using magneto-optic imaging nondestructive testing method of welding workpiece defect, combining Faraday electromagnetic induction principle, Faraday magnetically induced optical rotation effect, Brewster's law and computer machine vision, the polarized light source is studied. Design of polarized light path and control parameters of magneto-optic excitation mode. The polarization detector is used to detect the rotation change of polarized light caused by the change of magnetic field component in the position of welding defect by magnetized welding workpiece in excitation field, and the magneto-optical image of light and dark change is obtained. In the actual welding defect detection process, due to the change of magnetic field intensity and various other noises, the acquired magneto-optic image will still have the problem of degradation, the sharpness is poor, and the recognition degree is reduced. It is difficult to make a thorough analysis of the magneto-optic image of welded workpiece defects. Therefore, according to the characteristics of the detection environment of magneto-optic induction imaging, the restoration algorithm of magneto-optical image of welding defects is studied. By analyzing the characteristics of magneto-optic image of welding workpiece defects, using the principle of fuzzy set theory, using the improved continuous fuzzy enhancement algorithm to improve the discrimination, and taking high-strength steel as the detection object, the uneven magneto-optic imaging of the surface crack is studied. The deficiency of low differentiation between crack and melting zone. An adaptive fast edge detection algorithm is used to extract the texture features of welding crack images. The edge feature of crack magneto-optic image is extracted to improve the accuracy of welding crack detection and tracking process. In this paper, the image processing algorithm is used to identify and estimate the optimal scale of the weld magneto-optic image, and the extracted welding defect features are analyzed and trained according to the classification algorithm. The characteristic quantity of welding defect is constructed and the magneto-optic image of welding material surface defect is identified automatically. The experimental results show that the welding defect characteristics can be obtained by using magneto-optic imaging method, and the position, shape and category of weld defects can be identified by analyzing the fractal dimension of the image. At the same time, the effectiveness of extracting welding defect characteristics by principal component analysis (PCA) and using support vector machine (SVM) to identify defects in welding materials by magneto-optic imaging (MRI) is explored to improve the classification accuracy.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TG441.7

【相似文献】