作者：李茂, 杨录, 张艳花
作者单位：中北大学 电子测试技术国家重点实验室, 山西 太原 030051
关键词：超声缺陷信号;经验模态分解;主成分分析;特征提取
摘要：
针对非线性、非平稳超声缺陷信号的特征提取问题，提出一种经验模态分解（EMD）和主成分分析（PCA）相结合的缺陷信号特征提取方法。对缺陷信号进行EMD分解得到本征模态函数（IMF），根据能量比率累积选取IMF，平均截取傅里叶变换后的各模态频谱得到能表征原信号的特征向量集；构建PCA模型，特征向量集降维得到低维特征向量，该过程可降低缺陷信号分析数据的复杂度和冗余度，以BP神经网络为缺陷分类器对缺陷特征进行识别与分类。实验结果表明该方法具有可靠的识别与分类效果。
Research on feature extraction of ultrasonic flaw signal based on EMD and principal component analysis
LI Mao, YANG Lu, ZHANG Yanhua
National Key laboratory for Electronic Measurement Technology, North University of China, Taiyuan 030051, China
Abstract: Aiming at the feature extraction of non-linear and non-stationary ultrasonic flaw signal, a feature extraction method combining EMD and PCA was proposed. The flaw signal was decomposed to obtain IMF, and IMF was selected according to the energy ratio accumulation, and the spectrum of each modal after the Fourier transform was averaged to obtain the eigenvector of original signal. The PCA model was constructed and the dimension reduction of eigenvector set was achieved to obtain lower-dimension eigenvector, which reduced the data complexity and redundancy of the flaw signal analysis. Besides, BP neural network was used as the flaw classifier to identify and classify the defect features. The test results show that the proposed method has reliable flaw detection and classification effect on flaw signal.
Keywords: ultrasonic flaw signal;EMD;PCA;feature extraction
2018, 44(2): 118-121,133 收稿日期: 2017-03-19;收到修改稿日期: 2017-04-20
基金项目:
作者简介: 李茂(1993-),男,湖北仙桃市人,硕士研究生,专业方向为信息信号处理。
参考文献
[1] 蔡强福,陈鹏,韩德来. 基于小波包变换的电磁超声接收信号特征提取[J]. 仪表技术与传感器,2013(2):79-82.
[2] 刘深,张小蓟,牛奕龙,等. 基于IMF能量谱的水声信号特征提取与分类[J]. 计算机工程与应用,2014,50(3):203-206,226.
[3] 朱文龙,周建中,肖剑,等. 独立分量分析-经验模态分解特征提取在水电机组振动信号中的应用[J]. 中国电机工程学报,2013,33(29):95-101.
[4] 杨录,樊建平,张艳花. 一种基于EMD和神经网络的超声缺陷信号分类方法[J]. 中北大学学报(自然科学版),2012,33(5):598-602,616.
[5] 朱忠奎,陈祥芹,樊嶶,等. 基于降维的独立分量分析及信号特征成分检测[J]. 仪器仪表学报,2014,35(4):917-923.
[6] HUANG N E, SHEN Z, LONG S R, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Procedings Mathematical Physical & Engineering Sciences,1998,454(1971):903-995.
[7] 杨波,李世平. 一种基于GEP的EMD端点效应抑制方法[J]. 中国测试,2015,42(12):32-35.
[8] 李大中,赵杰. 基于EMD和GA-SVM的超声检测缺陷信号识别[J]. 中国测试,2016,42(1):102-106.
[9] 方健,李自品,彭辉,等. 基于主成分分析的BP神经网络的应用[J]. 变压器,2011,48(1):47-51.
[10] 叶永伟,刘志浩,黄利群. 基于PCA的汽车涂装线设备信号特征提取[J]. 仪器仪表学报,2011,32(10):2363-2370.
[11] 唐勇波,桂卫华,彭涛,等. PCA和KICA特征提取的变压器故障诊断模型[J]. 高电压技术,2014,40(2):557-563.
[12] MNASSRI B, ADEL E M E, ANANOU B, et al. Fault detection and diagnosis based on PCA and a new contribution plot[J]. IFAC Proceedings Volumes,2009,42(8):834-839.
