基于LBP和SVM的焊缝缺陷识别方法

Defect Identification Method of Weld Inspection Based on LBP and SVM

为保证焊接接头处于安全工作状态,对焊缝缺陷实施定量识别与分类,提出了基于局部二值模式(LBP)和支持向量机(SVM)的缺陷识别方法。首先采用局部二值模式LBP算法对焊缝的超声回波信号进行特征提取,并结合因子分子进行数据降维,降低高维特征集中的冗余数据,最后采用SVM模型实现缺陷的分类识别,并对影响SVM分类效果的核函数和超参数进行了优选。结果表明,高斯核函数在焊缝缺陷分类上的识别效果最好,当超参数C和特征向量ε分别为5.749 7和9.243 6,核函数的gamma参数为2.859 5时,模型最优,分类准确率为95%,分类效果优于常规时频域特征。研究结果可为焊缝缺陷的无损检测和评价提供实际参考。

In order to ensure that the welded joint is in a safe working state and carry out quantitative identification and classification of weld defects, a defect identification method based on local binary pattern(LBP) and support vector machine(SVM) is proposed. Firstly, the LBP algorithm is used to extract the features of the ultrasonic echo signal of the weld, and combined with the factor molecules to reduce the dimension of the data and reduce the redundant data in the high-dimensional feature set. Finally, the SVM model is used to realize the classification and recognition of defects, and the kernel function and super parameters affecting the classification effect of SVM are optimized. The results show that the recognition effect of Gaussian kernel function in weld defect classification is the best. When the super parameters C and ε are 5.749 7 and 9.243 6respectively, and the gamma parameter of kernel function is 2.859 5, the model is the best, the classification accuracy is 95%,and the classification effect is better than the conventional time-frequency domain features. The research results can provide practical reference for nondestructive testing and evaluation of weld defects.