基于PCA与SVM的焊缝缺陷信号分类方法

doi:10.16180/j.cnki.issn1007-7820.2020.05.002

摘要/Abstract

摘要：

为保证焊接结构处于安全工作状态,文中针对焊缝缺陷的分类问题提出一种结合主成分分析和支持向量机的焊缝特征导波缺陷信号的分类方法。该方法首先以缺陷回波的稀疏重构信号为基础提取了缺陷信号的特征参数矩阵,然后利用主成分分析法对参数矩阵进行降维优化以消除冗余信息。将优化的低维特征矩阵应用于支持向量机进行分类训练,并对比不同主成分的分类效果。最后选择核函数和相关参数来提高分类器的准确率。实验结果表明,该方法能够有效地对焊缝缺陷进行分类。

关键词: 焊缝, 回波信号, 缺陷分类, 特征导波, 主成分分析, 支持向量机

Abstract:

In order to ensure that the welding structure is in safe working state, a classification method combining PCA and SVM was proposed to classify the weld feature guided wave defect signals. Firstly, based on the sparse reconstructed signal of defect echo, the feature parameter matrix of defect signal was extracted, and the dimensionality of parameter matrix was optimized by PCA to eliminate redundant information. Then, the optimized low-dimensional feature matrix was applied to SVM for classification training, and the classification effect of different principal components was compared. Kernel function and related parameters were selected to improve the accuracy of the classifier. The experimental results showed that the method could effectively classify weld defects.

Key words: weld, echo signals, defect classification, feature guided wave, PCA, SVM

中图分类号:

TN911

万东燕,许桢英,杨卿,吴梦琪. 基于PCA与SVM的焊缝缺陷信号分类方法[J]. 电子科技, 2020, 33(5): 9-14.

WAN Dongyan,XU Zhenying,YANG Qing,WU Mengqi. A Method of Weld Defect Classification Based on PCA and SVM[J]. Electronic Science and Technology, 2020, 33(5): 9-14.

图/表 10

图1

图2

图3

表1

图4

图5

表2

图6

图7

表3

参考文献 16

[1]	Leinov E, Lowe M J S, Cawley P .Investigation of guided wave propagation and attenuation in pipe buried in sand[J]. Journal of Sound & Vibration, 2015,34(7):96-114.
[2]	何存富, 郑明方, 吕炎 , 等. 超声导波检测技术的发展、应用与挑战[J]. 仪器仪表学报, 2016,37(8):1713-1735.
	He Cunfu, Zheng Mingfang, Lü Yan , et al. Development,applications and challenges in ultrasonic guided waves testing technology[J]. Chinese Journal of Scientific Instrument, 2016,37(8):1713-1735.
[3]	Fan Z . Interaction of weld-guided waves with defects[J]. Ndt & E International, 2012,47(3):124-133.
[4]	杨川 . 焊缝特征导波模态分析方法的研究[D]. 镇江:江苏大学, 2017.
	Yang Chuan . Research on modes analysis of weld feature-guided wave[D]. Zhenjiang:Jiangsu University, 2017.
[5]	Ujjwal Kumar, Inderjeet Yadav, Shilpi Kumari , et al. Defect identification in friction stir welding using discrete wavelet analysis[J]. Advances in Engineering Software, 2015,85(7):43-50.
[6]	Luo H, Chen Q H, Doan N C N ,et al.Automated identification and characterization of clustered weld defects [C]. Alberta:IEEE International Conference on Advanced Intelligent Mechatronics (AIM) Banff, 2016.
[7]	董本志, 丁文雪 . 大数据环境下超声波焊缝缺陷识别方法的研究[J].计算机工程与应用, 2016(23):176-180.
	Dong Benzhi, Ding Wenxue . Research of ultrasonic weld defect identification method under big data environment[J].Computer Engineering & Applications, 2016(23):176-180.
[8]	王雒瑶, 高炜欣, 王欣 . 一种基于SVM及LE降维的X射线焊缝缺陷分类算法研究[J].西安石油大学学报(自然科学版), 2017(5):96-106.
	Wang Luoyao, Gao Weixin, Wang Xin . Research of X-ray weld defect classification algorithm based on SVM and LE dimensionality reduction[J].Journal of Xi’an Shiyou University(Natural Science Edition), 2017(5):96-106.
[9]	Murta R H F, Vieira Flávison de A, Santos V O , et al. Welding defect classification from simulated ultrasonic signals[J]. Journal of Nondestructive Evaluation, 2018,37(5):40-45.
[10]	Li Y, Cichocki A, Amari S . Analysis of sparse representation and blind source separation[J]. Neural Computation, 2014,16(6):1193-1234.
[11]	王星, 周一鹏, 田元荣 , 等. 基于改进遗传算法和Sin Chirplet原子的调频雷达信号稀疏分解[J]. 上海交通大学学报(自然版), 2017,51(9):1124-1130.
	Wang Xing, Zhou Yipeng, Tian Yuanrong , et al. Sparse decomposition for frequency modulation radar signal based on advanced genetic algorithm and sin-chirplet atom[J]. Journal of Shanghai Jiaotong University(Natural Science Edition), 2017,51(9):1124-1130.
[12]	Li J, Tao D . On preserving original variables in bayesian PCA with application to image analysis[J]. IEEE Transactions on Image Processing, 2012,21(12):4830-4843.
[13]	Yu Z H, Chin W L . Blind false data injection attack using PCA approximation method in smart grid[J]. IEEE Transactions on Smart Grid, 2015,6(3):1219-1226.
[14]	魏文博, 蔡红军 . 基于支持向量机的窄带雷达弹道导弹目标识别技术[J]. 电子科技, 2016,29(6):75-78.
	Wei Wenbo, Cai Hongjun . Narrowband radar ballistic missile target recognition technology based on SVM[J]. Electronic Science and Technology, 2016,29(6):75-78.
[15]	Wu J, Yang H . Linear regression-based efficient SVM learning for large-scale classification[J]. IEEE Transactions on Neural Networks & Learning Systems, 2017,26(10):2357-2369.
[16]	Balakrishna P, Raman S, Trafalis T B , et al. Support vector regression for determining the minimum zone sphericity[J]. International Journal of Advanced Manufacturing Technology, 2008,35(9-10):916-923.

PCA	PC1	PC2	PC3	PC4	PC5	PC6	PC7
特征值	16.06	10.39	8.41	5.73	0.84	0.08	0.02
贡献率	0.585 5	0.195 7	0.132 5	0.072 8	0.010 8	0.001 1	0.000 4

正确率	原始参数矩阵	1主成分	3主成分	5主成分
线性核	90%	76%	78%	83%
多项式核	96%	80%	87%	92%
高斯核	94%	88%	93%	98%
Sigmoid核	60%	33%	64%	60%

缺陷类型	识别个数	识别率	总识别率
孔洞	49/50	98%	98%
裂纹	48/50	96%
无缺陷	50/50	100%