光滑L0范数广义逆波束形成-论文.pdf

《光滑L0范数广义逆波束形成-论文.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、第35卷第6期仪器仪表学报Vol_35No．62014年6月ChineseJournalofScientificInstrumentJun．2014光滑LO范数广义逆波束形成术徐中明，黎术，贺岩松，张志飞，毛锦(1．重庆大学机械传动国家重点实验室重庆400044；2．重庆大学机械工程学院重庆400044)摘要：为改善Ll范数广义逆波束形成性能，提出了一种光滑IJo范数广义逆波束形成算法。该算法充分利用【D范数能充分反映单极子噪声源信号稀疏本质的特性，结合光滑IJD范数方法通过少数几次迭代即得到噪声源信号的稀疏解。首先，基于数值模拟的声源识别成像图以及声学半径曲线

2、探究其结果随声源频率、阵列形式的变化规律，结果表明光滑LD范数广义逆波束形成相比L1广义逆波束形成，在相同的空间分辨率情况下，需要的传感器测量值更少；在相同的声阵列形式下，光滑IJD范数广义逆波束形成的空间分辨率更高。最后，通过实验算例对单极子声源进行了声学成像，结果表明光滑IJ0范数广义逆波束形成相比其他波束形成，能更好地识别噪声源，进一步验证了该算法的有效性及实用性。关键词：空间分辨率；光滑IJD范数；广义逆波束形成；噪声源识别中图分类号：TB52文献标识码：A国家标准学科分类代码：140．20SmoothedLOnormgeneralizedinverse

3、beamformingXuZhongming，LiShu，HeYansong，ZhangZhifei，MaoJin(．SlateKeyLaboratoryofMechanwalTransmission，Chongqing，Chongqing4OOO44，China；2．SchoolofMechanicalEngineering，ChongqingUniversity，Chongqing4O0O44，China)Abstract：ToimprovetheperformanceofL1normgeneralizedinversebeamforming，anewnoi

4、sesourcedetectionalgorithmwasdevelopedbasedonSL0一GIB(smoothedL0normgeneralizedinversebeamforming)．Thismethodadequatelyutilizesthechar—acteristicthatL0normcanfullyreflectthesparsitynatureofmonopolenoisesourcesignal；andcanobtainthesparsitysolutionofthenoisesourcewithafewtimesofiteratio

5、nsbycombiningsmoothedL0normmethod．Firstly，thevaryingrulesofthesoundsourceidentificationresultswithsignal~equencyandarraytypeareexploredbasedonnumericalsimulatedsoundsourceidenti—ficationimagingmapsandstandarddeviationcurvesoftheacousticradius．ThesmoothedLOnormGIBiscomparedwithL1一GIB(

6、L1normgeneralizedinversebeamforming)．TheresultsindicatethatunderthesamespatialresolutionthesmoothedL0normgeneralizedinversebeamformingrequiresfewerpressurehydrophonemeasurementvalues；andunderthesameacousticar—raytype，theSL0一GIBhashigherspatialresolution．Finally，asatestexample，theacou

7、sticimagingofmonopolenoisesourcewasconducted．ExperimentanddataanalysisresultsshowthatthesmoothedL0normgeneralizedinversebeamformingcanidentifythenoisesourcepositionbetter，andtheresolutionoftheproposedalgorithmissubstantiallyhigherthanthatofotheral—gorithms，whichfurtherverifiestheeffe

8、ctivenessand