作者：彭宁伟, 张颖, 王雪琴, 赵鹏程

作者单位：常州大学环境与安全工程学院，江苏 常州 213164

关键词：焊接冷裂纹;声发射技术;变分模态分解;能量熵;GA-SVM

摘要：

针对低合金高强钢焊接过程中存在焊接工艺不规范导致的延迟冷裂纹，提出一种监测与分类识别方法。首先利用声发射 (AE) 技术对焊后试件进行持续监测，之后根据出现冷裂纹试件的监测数据选取冷裂纹声发射信号中的起裂信号和氢聚信号，并对两者进行基于变分模态分解(VMD)的能量熵特征提取，最后利用遗传算法(GA)将传统支持向量机(SVM)进行改进后对信号进行分类识别，并结合传统支持向量机(SVM)的识别分类结果进行对比。同时为了验证VMD能量熵相较于其他能量熵在特征提取上的精准性，将提取冷裂纹声发射信号的EMD能量熵和CEEMDAN能量熵进行分类识别效果对比。结果表明，利用VMD能量熵作为焊接冷裂纹声发射信号的识别特征相较于其他能量熵特征识别精度更高，且随着支持向量机的优化识别精度会进一步提高到95%。

Research on classification method of welding cold crack’s AE signals based on VMD energy entropy and GA-SVM
PENG Ningwei, ZHANG Ying, WANG Xueqin, ZHAO Pengcheng
School of Environmental & Safety Engineering, Changzhou University, Changzhou 213164, China
Abstract: Aiming at the delayed cold crack caused by non-standard welding process in the welding process of low-alloy high-strength steel, a monitoring and classification method is proposed. First, acoustic emission (AE) technology is used to continuously monitor the post-welding specimen. Then, crack initiation signal and hydrogen aggregation signal in the AE signal of cold crack are selected according to the monitoring data of the specimen with cold crack, and energy entropy feature extraction based on variational mode decomposition (VMD) is carried out for both. Finally, genetic algorithm (GA) is used to improve the traditional support vector machine (SVM) to classify and recognize the signal, and the recognition and classification results of the traditional SVM are compared. At the same time, in order to verify the accuracy of VMD energy entropy in feature extraction, the EMD energy entropy and CEEMDAN energy entropy of cold crack acoustic emission signal extraction were compared for classification and recognition effect. The results show that when the VMD energy entropy is used as the identification feature of the acoustic emission signal of welding cold crack, the recognition accuracy is higher than other energy entropy features, and the recognition accuracy will further increase to 95% with the optimization of support vector machine.
Keywords: welding cold crack;acoustic emission technology;variational mode decomposition;energy entropy;GA-SVM
2024, 50(5):47-53  收稿日期: 2022-04-29;收到修改稿日期: 2022-06-17
基金项目: 中国石油天然气集团有限公司常州大学创新联合体科技合作项目（KC20210301）
作者简介: 彭宁伟(1997-),男,安徽亳州市人,硕士研究生,专业方向为特种设备健康监测及智能诊断。
参考文献
[1] 林厚起, 孙青燕. 压力容器焊接过程中的常见问题分析与防范方法[J]. 流体测量与控制, 2022, 3(3): 43-46
[2] 房海基, 吕波, 张艳喜, 等. 焊接过程声信号在线检测技术现状与展望[J]. 精密成形工程, 2022, 14(1): 165-172
FANG H J, LV B, ZHANG Y X, et al. Status and prospect of on-line acoustic signal detection technology in welding[J]. Journal of Netshape Forming Engineering, 2022, 14(1): 165-172
[3] 蒋宇涵, 华春蓉, 董大伟, 等. 基于优化VMD的车轴裂纹和车轮扁疤故障诊断[J]. 噪声与振动控制, 2021, 41(6): 71-76
JIANG H Y, HUA C R, DONG D W, et al. Fault diagnosis of axle cracks and wheel flats based on optimized VMD[J]. Noise and Vibration Control, 2021, 41(6): 71-76
[4] 蒋宇涵, 华春蓉, 熊丽波, 等. 含裂纹轮对的振动特性分析及裂纹参数识别[J]. 中国测试, 2022, 48(1): 154-159
JIANG H Y, HUA C R, XIONG L B, et al. Vibration characteristics analysis and crack parameters identification of cracked wheelset[J]. China Measurement & Test, 2022, 48(1): 154-159
[5] 唐贵基, 刘尚坤. 基于VMD和谱峭度的滚动轴承早期故障诊断方法[J]. 中国测试, 2017, 43(9): 112-117
TANG G J, LIU S K. Incipient fault diagnosis method for rolling bearing based on VMD and spectral kurtosis[J]. China Measurement & Test, 2017, 43(9): 112-117
[6] 丁世飞, 齐丙娟, 谭红艳. 支持向量机理论与算法研究综述[J]. 电子科技大学学报, 2011, 40(1): 2-10
DING S F, QI B J, TAN H Y. An overview on theory and algorithm of support vector machines[J]. Journal of University of Electronic Science and Technology of China, 2011, 40(1): 2-10
[7] 罗光宇, 杨光明. 基于MIV分析的GA-BP神经网络闸门健康诊断[J]. 水电能源科学, 2021, 39(11): 203-206
[8] DRAGOMIRETSKIY K, ZOSSO D. Variational mode decomposition[J]. IEEE Tranactions on Signal Processing, 2014, 62(3): 531-544
[9] 邓飞跃. 基于自适应谐波小波和能量熵的转子系统故障诊断研究[J]. 中国测试, 2016, 42(8): 105-107
[10] CORTES C, VAPNIK V. Support-vector networks[J]. Machine Learning, 1995, 20(3)273-297.
[11] 周俊鹏. SPV490Q钢焊接冷裂纹声磁特性及实验方法研究[D]. 大庆: 东北石油大学, 2017.
ZHOU J P. Study on the acoustic-magnetic properties and experimental method of spv490q steel welding cold crack[D]. Daqing: Northeast Petroleum University, 2017.
[12] 成勇. 焊接裂纹声发射信号特性实验研究[D]. 湘潭: 湖南科技大学, 2016.
CHENG Y. Experimental study of welding crack acoustic emission signal characteristics[D]. Xiangtan: Hunan University of Science and Technology, 2016.
[13] 肖倩, 王建辉, 方晓柯, 等. 一种基于互相关函数的小波系数相关阈值去噪方法[J]. 东北大学学报(自然科学版), 2011, 32(3): 318-321
XIAO Q, WANG J H, FANG X K, et al. A Wavelet coefficient threshold denoising method based on a cross-correlation function[J]. Journal of Northeastern University(Natural Science), 2011, 32(3): 318-321