依据声信号频率分布和复杂度的供水管道泄漏辨识
- 海之魂
-
0 次阅读
-
0 次下载
-
2020-02-26 12:59:03
文档简介:
第35卷第6期2014年6月仪器仪表学报ChineseJournalofScientificInstrumentVo1.35No.6Jun.2014依据声信号频率分布和复杂度的供水管道泄漏辨识术文玉梅,张雪园,文静,甄锦鹏,王凯(重庆大学光电工程学院传感器与仪器研究中心重庆400044)摘要:为了实现不同工况下的供水管道泄漏检测,提出采用谱宽参数和近似熵作为特征量进行管道泄漏辨识的方法。根据泄漏声信号与噪声信号在频谱内能量分布的差异,采用谱宽参数作为特征量辨识泄漏信号与宽带噪声、窄带噪声;对于与泄漏信号在频谱内的能量分布相当的固定干扰噪声可能引起的误判,则利用泄漏信号和固定干扰噪声的随机性差异,引进近似熵复杂度作为特征量进行辨识。将经去噪处理后检测信号的谱宽参数和近似熵同时作为支持向量机的输入,辨识管道泄漏。对工程实际检测的信号进行处理,结果表明泄漏辨识的准确率大于93%,采用该方法能够有效辨识供水管道的泄漏信号与环境中存在的多种非固定干扰噪声及『闾定干扰噪声。关键词:泄漏辨识;能谱宽;近似熵;支持向量机中图分类号:TN911.7TH86文献标识码:A国家标准学科分类代码:510.50IdentificationofwaterpipelineleakagebasedonacousticsignalfrequencydistributionandcomplexityWenYumei,ZhangXueyuan,WenJing,ZhenJinpeng,WangKai(ResearchCenteroJSensorsandInstruments,Schoolof@to—electronicEngineering,ChongqingUniversity,Chongqing400044,China)Abstract:Inordertoachieveeffectiveleakdetectionofwaterpipelineunderdifferentconditions,thispaperproposesamethodtoidenti—fypipelineleaksignalsusingtheenergyspectrumwidthparameterandapproximateentropyasthefeatureparameters.Accordingtothedifferenceoftheenergydistributioninfrequencyspectrumbetweenleaksignalandnoisesignal,theenergyspectrumwidthisusedasfeatureparametertoidentifytheleaksignalsandnarrowbandnoises&widebandnoises.Todealwiththemisjudgmentcausedbythefixedinterferencenoisesthathavesimilarenergydistributioninthe~equeneyspectrumoftheleaksignals,thestochasticdiferencebe—tweenthesetwokindsofsignalsisusedandtheapproximateentropycomplexityisintroducedasthefeaturetoidentifytheleaksignalsandfixedinterferencenoises.Thesuppo~vectormachineisdevelopedasaclassifier,theenergyspectrumwidthandapproximateentro—pyofthedenoiseddetectionsignalaretakenasthenetworkinputsofthesupportvectormachine,andthepipelineleakageisidentified.Thedetectionsignalsinengineeringapplicationwereprocessedwiththeproposedmethod.Theprocessingresultsshowthatthecorrectleakagedetectionratioisgreaterthan93%,andthismethodcaneffectivelyidentifytheleaksignalsfromvariousnon—fixedinterferencenoisesandfixedinterferencenoisesinwaterpipeline.Keywords:leakdetection;energyspectrum
评论
发表评论