基于X射线实时成像的铝合金激光焊接缺陷识别技术研究

发布时间：2018-11-04 17:09

【摘要】：铝合金激光焊接工艺广泛应用于民用飞机壁板等航空航天结构件的制造。在激光焊接缺陷检测中,传统X射线人工检测结果受胶片本身局限、主观人为因素影响较大。随着新一代X射线实时成像技术以及图像模式识别技术的飞速发展,使得基于X射线数字图像的计算机辅助识别铝合金激光焊接缺陷成为可能。本文针对铝合金激光焊件X射线实时成像技术、图像预处理技术以及缺陷提取与识别技术的应用开展研究,初步实现了铝合金激光焊接缺陷的自动提取与识别。首先,在合理选择硬件系统配置的基础上,设计并搭建了面向铝合金T型接头激光焊接件的X射线实时成像系统。通过搭建的成像系统,对铝合金T型接头激光焊件的X射线实时成像检测技术进行了系统的研究,并得到了最佳的成像工艺参数:管电压60kV,管电流0.3mA,焦距350mm,放大倍数2.4。其次,针对X射线原始图像的特点,通过图像灰度转换、图像降噪以及图像模糊增强的预处理功能对焊件X射线图像进行质量改善处理。其中,通过对单一降噪方法的结合使用,提高了对混合噪声的去除效果。之后,对传统模糊增强算法的改进应用增强了图像的对比度,为后续焊缝提取与缺陷分割提供了质量良好的X射线图像。再次,通过进一步对X射线图像列灰度曲线的分析,采用曲线拟合和灰度差分判定的方法完成了对复杂背景条件下焊缝区域的提取。同时,通过自适应形态学滤波算法模拟出焊缝背景图像,再经差影检测和迭代阈值分割算法的处理,实现了焊缝中缺陷的分割。之后,利用轮廓提取与种子填充算法完成了对缺陷的提取。最后,在缺陷区域标记的基础上,完成了各类特征参数的提取与计算。根据缺陷的特征参数及其他X射线图像影像特征,设计并开发了基于正向模糊推理的缺陷识别与分类专家系统。其中,专家系统知识库中的经验知识以规则的形式存在,用户可对规则进行修改、添加与删除等操作。同时,根据相关标准实现了铝合金激光焊件圆形缺陷的评定。本文针对铝合金T接头激光焊件的特点,通过X射线实时成像实验和计算机编程图像处理仿真实验,完成了对铝合金激光焊件的X射线实时成像、X射线图像处理以及图像缺陷识别技术的研究,为铝合金激光焊接结构缺陷的无损检测提供了新的途径。
[Abstract]:The laser welding technology of aluminum alloy is widely used in the manufacture of aerospace structures such as civil aircraft wall panels. In laser welding defect detection, the traditional X-ray manual detection results are limited by the film itself, subjective human factors have a greater impact. With the rapid development of the new generation of X-ray real-time imaging technology and image pattern recognition technology, it is possible to identify the defects of aluminum alloy laser welding based on X-ray digital image. In this paper, the application of X-ray real-time imaging technology, image preprocessing technology and defect extraction and recognition technology for laser welding of aluminum alloy are studied, and the automatic extraction and recognition of aluminum alloy laser welding defects are preliminarily realized. Firstly, based on the reasonable selection of hardware system configuration, a real-time X-ray imaging system for laser welding of aluminum alloy T-joints is designed and built. By using the imaging system, the X-ray real time imaging technology for laser welding of aluminum alloy T joint is studied systematically. The optimum imaging parameters are obtained as follows: tube voltage 60 kV, tube current 0.3 Ma, focal length 350 mm, tube voltage 60 kV, tube current 0.3 Ma, focal length 350 mm, tube voltage 60 kV, tube current 0.3 Ma, focal length 350 mm. The magnification is 2.4. Secondly, according to the characteristics of the original X-ray image, the image quality is improved by image gray conversion, image denoising and image fuzzy enhancement. Among them, the combined use of a single noise reduction method improves the removal effect of mixed noise. Then, the improvement of the traditional fuzzy enhancement algorithm enhances the contrast of the image, and provides a good quality X-ray image for the subsequent weld extraction and defect segmentation. Thirdly, through the further analysis of X-ray image column gray curve, the method of curve fitting and gray difference judgment is used to complete the extraction of weld area under complex background conditions. At the same time, the weld seam background image is simulated by the adaptive morphological filtering algorithm, and then the defect segmentation in the weld is realized by differential image detection and iterative threshold segmentation algorithm. After that, the defect extraction is accomplished by contour extraction and seed filling algorithm. Finally, on the basis of defect area marking, the extraction and calculation of all kinds of feature parameters are completed. A defect recognition and classification expert system based on forward fuzzy reasoning is designed and developed according to the feature parameters of defects and other X-ray image features. Among them, the empirical knowledge in expert system knowledge base exists in the form of rules. Users can modify, add and delete rules. At the same time, the circular defects of laser welding parts of aluminum alloy are evaluated according to the relevant standards. In this paper, according to the characteristics of laser welding parts of aluminum alloy T-joints, X-ray real-time imaging of aluminum alloy laser welds has been completed by means of X-ray real-time imaging experiment and computer programming image processing simulation experiment. The research of X-ray image processing and image defect recognition provides a new way for nondestructive detection of aluminum alloy laser welding structural defects.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TG441.7

【参考文献】