基于ICEEMDAN-ICA的焊缝信号去噪算法.pdfVIP

HotWorkingTechnology2022,Vol.51,No.l

•澤接技术・

基于ICEEMDAN-ICA的焊缝信号去噪算法

翟永杰，杨旭

(华北电力大学自动化系，河北保定071003)

摘要：随着视觉传感器在焊缝识别领域的应用，基于线阵电荷耦合元件(CCD)的焊缝识别技术以其成本低、实时

性好等优势受到广泛关注，但其准确度往往与滤波技术密切相关。为此，提出基于ICEEMDAN-ICA的焊缝信号去噪算

法，首先利用线阵CCD釆集焊板的焊缝灰度图像，然后通过改进完备经验模态分解(ICEEMDAN)和独立成分分析

(ICA)算法得到焊缝信号的多个源信号，通过模糊爛阈值判据识别并剔除其中的噪声源信号，最后重构得到焊缝滤波

信号。实验结果表明，该算法能更好去除焊接信号的髙频噪声，保留其中的有效信息。

关键词：焊缝识别；线阵CCD；改进完备经验模态分解(ICEEMDAN)；独立成分分析(ICA)；模糊嫡

DOI:10.14158/j.cnki.1001-3814.

中图分类号:TG409文献标识码：A文章编号：1001-3814(2022)01-0096-07

WelSignalDenoisingAlgorithmforWeldsBaseonICEEMDAN-ICA

ZHAIYongjie,YANGXu

(DepartmentofAutomation,NorthChinaElectricityPowerUniversity,Baoding071003,China)

Abstract：Withtheapplicationofvisualsensorinthefieldofweldingseamidentification,theweldingseam

identificationtechnologybasedonlineararraychargecoupleddevices(CCD)hasattractedwideattentionduetoits

advantagessuchaslocostandgoodreal-timeperformance,butitsaccuracyisoftencloselyrelatedtothefiltering

technology.Therefore,theweldsignaldenoisingalgorithmbasedonICEEMDAN-ICAwasproposed.Firstly,theweldgrey

imagewasacquiredbyusingthelinearCCDfirst,andthenmultiplesourceofweldsignalswasobtainedbybyimproved

complementaryensembleempiricalmodedecompositionwithadaptivenoiseandindependentcomponentanalysisalgorithm,

thenoisesignalwasidentifiedandeliminatedthroughthefuzzyentropythresholdcriterion,finallytheweldfilteringsignal

wasreconstructed.Theexperimentalresultsshothatthisalgorithmcanremovethehighfrequencynoiseofweldingsignal

andretaint