作者:刘文芳1,2, 韩军2, 刘艳锋2, 龙晋桓2, 林靖翔3, 陈尚群3
作者单位:1. 中北大学电气与控制工程学院,山西 太原 030051;
2. 中国科学院福建物质结构研究所 泉州装备制造研究所,福建 泉州 362000;
3. 泉州德源轴承实业有限公司,福建 泉州 362000
关键词:超声检测;近表面缺陷;缺陷回波;支持向量机
摘要:
在超声脉冲回波检测技术中,由表面回波产生的“死区”会隐藏近表面缺陷的回波信号,对于近表面缺陷的实际检测造成一定的干扰。针对上述问题,提出一种将数字信号处理与支持向量机相结合的方法,通过识别缺陷一次回波和二次回波,实现对近表面缺陷位置的准确计算。实验中,通过垂直入射超声脉冲回波法采集轴承内圈的A波信号,使用支持向量机分类法对缺陷的一次回波和二次回波数据进行训练、测试和分类。实验结果表明,该文所提方法能有效对缺陷一次回波和二次回波信号进行识别和对缺陷实际位置的预测,分类的平均准确率可达95.22%,近表面缺陷位置的预测误差绝对值在0.2 mm以内。
Research on ultrasonic detection method of metal near-surface defect based on machine learning
LIU Wenfang1,2, HAN Jun2, LIU Yanfeng2, LONG Jinhuan2, LIN Jingxiang3, CHEN Shangqun3
1. School of Electrical and Control Engineering, North University of China, Taiyuan 030051, China;
2. Quanzhou Institute of Equipment Manufacturing, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Quanzhou 362000, China;
3. Quanzhou Deyuan Bearing Industry Co., Ltd., Quanzhou 362000, China
Abstract: The ultrasonic pulse echo technique can be used for the defect detection. However, the “dead zone” created by the front-wall echo may overlap the defect echo, which interferes the detection of near-surface defect. In this paper, a method combining digital signal processing with support vector machine (SVM) was proposed to identify primary-echo and secondary-echo of the defects, and the accurate calculation of near surface detect position could be achieved finally. In the experiment, A-wave signals of the bearing have been collected by the normal-incidence immersion measurement, and the primary-echo and secondary-echo data of defects have been trained, tested and classified by support vector machine. The experimental results revealed that the proposed method could effectively identify the primary and secondary echo signals and prediction the actual position of defects. The average prediction accuracy reached 95.22%, and the absolute value of prediction error of near-surface defect position was less than 0.2 mm.
Keywords: ultrasonic detection;near-surface defect;defect echo;support vector machine
2022, 48(1):46-52 收稿日期: 2021-01-14;收到修改稿日期: 2021-02-26
基金项目: 中国科学院对外重点合作项目(121835KYSB20180062);福建省科技计划项目(2018T3007,2019T3025);泉州市科技计划项目(2019STS04,2019STS07)
作者简介: 刘文芳(1996-),女,山东烟台市人,硕士研究生,专业方向为超声检测及超声无损评估
参考文献
[1] SCHNBAUER B M, YANASE K, ENDO M. The Influence of various types of small defects on the threshold behaviour of precipitation-hardened 17-4PH stainless steel[J]. Theoretical and Applied Fracture Mechanics, 2016, 87(2): 35-49
[2] HAN X, HUA L, ZHOU G, et al. A new cylindrical ring rolling technology for manufacturing thin-walled cylindrical ring[J]. International Journal of Mechanical Sciences, 2014, 81(4): 95-108
[3] 薛光辉, 刘昊, 何毛宁. 基于超声波的压接质量检测方法研究[J]. 中国测试, 2020, 46(6): 39-43
[4] 陈志恒, 罗文斌, 常俊杰, 等. 基于EMD的神经网络空耦超声储油罐液位检测[J]. 中国测试, 2021, 47(1): 9-14
[5] FRITSCH C, VECA A. Detecting small flaws near the interface in pulse-echo[J]. Ultrasonics, 2004, 42(1-9): 797-801
[6] LU X M, REID J M, SOETANTO K, et al. Cepstrum technique for multilayer structure characterization[J]. IEEE Ultrasonics Symposium. IEEE, 1990, 3(12): 1571-1574
[7] DRAI R, SELLIDJ F, KHELIL M, et al. Elaboration of some signal processing algorithms in ultrasonic techniques: application to materials NDT[J]. Ultrasonics, 2000, 38(1-8): 503-507
[8] GUAN S, WANG X, HUA L, et al. Quantitative ultrasonic testing for near-surface defects of large ring forgings using feature extraction and GA-SVM[J]. Applied Acoustics, 2021, 173(2): 107714
[9] LI M, LI X, GAO C, et al. Acoustic microscopy signal processing method for detecting near-surface defects in metal materials[J]. NDT & E international, 2019, 103(4): 130-144
[10] HUANG Y, TURNER JA, SONG Y, et al. Enhanced ultrasonic detection of nearsurface flaws using transverse-wave backscatter[J]. Ultrasonics, 2019, 98(9): 20-27
[11] KRIZHEVSKY, ALEX, SUTSKEVER, et al. Imagenet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90
[12] 李萍, 宋波, 毛捷, 等. 深度学习在超声检测缺陷识别中的应用与发展[J]. 应用声学, 2019, 38(3): 458-464
[13] 薛忠军, 张建龙, 卫文哲, 等. 基于小波包分析与支持向量机的桩身缺陷严重程度识别方法[J]. 公路交通科技(应用技术版), 2018(11): 216-220
[14] 窦希杰, 王世博, 谢洋, 等. 基于IMF能量矩和SVM的煤矸识别[J]. 振动与冲击, 2020, 39(24): 39-45
[15] ISA D, RAJKUMAR R. Pipeline defect prediction using support vector machines[J]. Applied Artificial Intelligence, 2009, 23(8): 758-771