資源描述:
《time frequency representation based on an adaptive short time fourier transform外語英文電子書》由會員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在教育資源-天天文庫。
1、5118IEEETRANSACTIONSONSIGNALPROCESSING,VOL.58,NO.10,OCTOBER2010Time-FrequencyRepresentationBasedonanAdaptiveShort-TimeFourierTransformJingangZhongandYuHuangAbstract—Inthispaper,anewconcisealgorithmabouttime-fre-FortheconventionalTFRalgorithms,short-timeFourierquencyrepresentation(
2、TFR)basedonanadaptiveshort-timetransform(STFT),wavelettransform(WT),andWigner–VilleFouriertransform(ASTFT)ispresented.Inthisalgorithm,thedistribution(WD)arepopularchoices.However,thewindowanalysiswindowwidthisequaltothelocalstationarylengthwhichwidthofSTFTisinvariable,soitsuffersp
3、oortime-frequencyismeasuredbytheinstantaneousfrequencygradient(IFG)ofthesignal.Andtheinstantaneousfrequency(IF)ofthesignalisresolution.TheWTfailsbecauseitswindowdilationruleisnotobtainedbydetectingtheridgeofwavelettransform(WT).Thesignal-dependent(explainedafterwardsbythe?rstexamp
4、leASTFTprovidesmuchbetterperformancethanconventionalTFRofnumericalsimulation).TheStransform[2]stillinheritsalgorithms.Furthermore,thealgorithmissimplerandmorecom-thelimitationsofWT.TheWDmakes?neTFRinsomecaseputationalef?cientthansomeofotheradaptiveTFRalgorithmsofmonocomponentsigna
5、lanalysis.However,itintroducesproposedpreviously.Severalexamplesarepresentedtoillustrateitsbehaviorondifferentkindsofsignalsanddemonstrateitscross-termsbetweenmultiplesignalcomponents.Itmakesvalidity.interpretationoftheWDdif?cultinmanypracticalapplica-IndexTerms—Adaptivesignalproc
6、essing,Fouriertransform,tions.ThecrossWD(XWD)[3]waspresentedtoremedytime-frequencyanalysis.thede?ciencyofWD.However,thismethodrequiresiterantcomputation.Inrecenttwodecades,ithadbecomemoreandmoreap-I.INTRODUCTIONparentthatagoodTFRalgorithmshouldbeasignal-dependentone,andthenadaptiv
7、eTFRalgorithmswereproposed.SomeLASSICALFouriertransform(FT)isastandardtoolforoftheresearchersbelievethatnosinglekernelcangiveade-Cspectrumanalysisofstationarysignalbecauseitrevealsquateperformanceforalargeclassofsignals.Oneofthemde-theoverallfrequencycontent.However,foranonstation
8、ary?nedoptimalradialGaussiankerne