管道缺陷多特征融合提取方法研究.pdfVIP

机械工程师

MECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEERMECHANICALENGINEER

MECHANICALENGINEER

管道缺陷多特征融合提取方法研究

都腾飞，李仲树，谢雪松，李泽军，刘新

（江苏长江智能制造研究院有限责任公司,江苏常州213164）

摘要：

针对管道缺陷的风险评估，预设了工程实际中常见的管道凹坑、孔洞和裂纹典型缺陷特征，构建了实验测试平台，

传感器采集信号。针对信号非线性非平稳的处理问题，提出一种基于ICEEMDAN-ICA的管道缺陷多特征融合信号处理方

法，对管道的孔洞、凹坑和裂纹缺陷进行多特征提取，提取样本熵、近似熵和奇异谱熵进行特征融合，通过ELM分类器进行

模式识别。研究结果表明，特征融合的方法大大提高了识别的准确率，达到了95%。

关键词：

故障诊断；往复泵；信号处理；特征提取；模式识别

中图分类号:TE973.6；TP391.41文献标志码:粤文章编号：员园园圆原圆猿猿猿（2021）02原园117原园5

ResearchonMulti-featureFusionExtractionMethodofPipelineDefects

DUTengfei,LIZhongshu,XIEXuesong,LIZejun,LIUXin

(JiangsuChangjiangIntelligentManufacturingResearchInstituteCo.,Ltd.,Changzhou213164,China)

Abstract:Fortheriskassessmentofpipelinedefects,thetypicaldefectcharacteristicsofpipelinepits,holesandcracks

thatarecommoninengineeringpracticearepreset,anexperimentaltestplatformisbuilt,andthesensorcollectssignals.

Aim