基于X射线检测的旋挖钻机桅杆焊缝缺陷识别

Identification of rotary drilling rig mast weld defects based on X-ray inspection

  • 摘要:
    目的 旋挖钻机作为基础工程建设的核心设备,其桅杆焊缝在极端工况下易出现应力集中与缺陷,直接影响施工安全与设备寿命。为了实现对旋挖钻机桅杆焊缝焊接质量的判定,利用X射线试验,深入研究了旋挖钻机桅杆焊缝的缺陷识别。
    方法 首先,通过简化桅杆结构模型并分析其在不同工况下的应力分布与变形,确定了大圆盘焊缝作为最危险位置进行X射线检测。对比了多种图像滤波和增强方法,确定了基于SUNet滤波和线性增强的预处理算法,显著提升了图像质量。进一步,通过研究焊缝图像的阈值分割和边缘检测技术,将Otsu’s阈值法和Canny算子相结合,有效提取了图像关键特征。在此基础上,分析了常见焊缝缺陷,构建包含 24 407 张 224×224 像素图像的数据集,涵盖裂纹、气孔、未焊透及无缺陷四类样本,按比例划分为训练集、验证集与测试集。设计残差识别网络,结合滑窗技术,提出了一种适应多种图像尺寸的缺陷识别算法。
    结果 该算法在缺陷识别中可达到精确度0.977 89,准确度0.981 88,召回率0.981 86,F1参数0.979 33,在不同尺寸的焊缝X射线检测图像中达到89%的准确率,并在旋挖钻机射线检测图像中成功应用。
    结论 该算法在旋挖钻机桅杆焊缝缺陷检测中表现出高适应性和精度,达到了良好的检测效果。为旋挖钻机焊缝缺陷提供了高效、高精度的无损检测方案,有效提升设备运维安全性与可靠性,对基础工程质量保障具有重要意义。

     

    Abstract: Objective As a core equipment in infrastructure construction, the mast welds of rotary drilling rigs are prone to stress concentration and defects under extreme working conditions, which directly affects construction safety and equipment service life. To realize the determination of the welding quality of rotary drilling rig mast welds, this paper uses X-ray tests to conduct an in-depth study on the defect identification of rotary drilling rig mast welds. Methods Firstly, by simplifying the mast structure model and analyzing its stress distribution and deformation under different working conditions, the large disc weld is determined as the most dangerous location for X-ray inspection. Various image filtering and enhancement methods are compared, and a preprocessing algorithm based on SUNet filtering and linear enhancement is determined, which significantly improves image quality. Further, by studying the threshold segmentation and edge detection technologies of weld images, Otsu’s threshold method and Canny operator are combined to effectively extract key image features. On this basis, common weld defects are analyzed, and a dataset containing 24 407 images of 224×224 pixels is constructed, covering four types of samples, cracks, pores, incomplete penetration, and non-defective ones, which are divided into a training set, a validation set, and a test set in proportion. A residual recognition network is designed, and combined with sliding window technology, a defect identification algorithm suitable for various image sizes is proposed. Results This algorithm can achieve a precision of 0.977 89, an accuracy of 0.981 88, a recall of 0.98186, and a F1 parameter of 0.979 33 in defect identification, reaching an accuracy rate of 89% in weld X-ray inspection images of different sizes, and has been successfully applied in rotary drilling rig X-ray inspection images. Conclusion The algorithm shows high adaptability and precision in the defect detection of rotary drilling rig mast welds, achieving good detection results. It provides an efficient and high-precision non-destructive testing scheme for rotary drilling rig weld defects, effectively improving the safety and reliability of equipment operation and maintenance, and is of great significance for the quality assurance of infrastructure projects.

     

/

返回文章
返回