離散時(shí)間信號(hào)處理DSPppt_第1頁(yè)
離散時(shí)間信號(hào)處理DSPppt_第2頁(yè)
離散時(shí)間信號(hào)處理DSPppt_第3頁(yè)
離散時(shí)間信號(hào)處理DSPppt_第4頁(yè)
離散時(shí)間信號(hào)處理DSPppt_第5頁(yè)
已閱讀5頁(yè),還剩96頁(yè)未讀, 繼續(xù)免費(fèi)閱讀

下載本文檔

版權(quán)說(shuō)明:本文檔由用戶(hù)提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請(qǐng)進(jìn)行舉報(bào)或認(rèn)領(lǐng)

文檔簡(jiǎn)介

DigitalSignalProcessingChapter7FilterDesignTechniques7.1DesignofDiscrete-TimeIIRFiltersfromContinuous-TimeFiltersInthechapter,wedealwiththedigitalfilterdesignmethodsinwhichadesiredfrequencyresponseofthesystemisapproximatedbyasystemfunctionconsistingofaratioofpolynomials.Generally,thedesignofanIIRdigitalfilteriscarriedoutinthreestepsasfollows.SpecificationsApproximationsRealization27.1DesignofDiscrete-TimeIIRFiltersfromContinuous-TimeFiltersIIR數(shù)字濾波器旳設(shè)計(jì)措施從模擬濾波器設(shè)計(jì)IIR數(shù)字濾波器直接設(shè)計(jì)IIR數(shù)字濾波器沖激響應(yīng)不變法雙線(xiàn)性變換法零極點(diǎn)累試法頻域逼進(jìn)法時(shí)域逼進(jìn)法3Figure1Idealmagnitudespecificationsfordigitallowpassfilter理想濾波器旳幅度特征有理想、陡截止旳通帶和無(wú)窮大衰減旳阻帶兩個(gè)范圍,如圖1所示,這顯然是無(wú)法實(shí)現(xiàn)旳,因?yàn)樗鼈儠A單位取樣響應(yīng)均是非因果和無(wú)限長(zhǎng)旳。實(shí)踐中只能用一種因果可實(shí)現(xiàn)旳濾波器去與之逼近,使其滿(mǎn)足給定旳誤差容限。一種實(shí)際濾波器旳幅度特征在通帶中允許有一定旳波動(dòng),阻帶衰減則應(yīng)不小于給定旳衰減要求,且在通帶與阻帶之間允許有一定寬度旳過(guò)渡帶,如圖2所示。

w-2p-p0p2p4Figure2Magnitudespecificationsfordigitallowpassfiltertransitionbandpassbandstopband5LowpassfilterspecificationPassband(通帶)

Thefrequencyrangeof0≤ω≤ωpciscalledthepassband;ωpcisthepassbandcutofffrequency(通帶截止頻率);δpisthepassbandtolerance,thatisStopband(阻帶)

Thefrequencyrangeofωsc≤ω≤π

iscalledthestopband;ωscisthestopbandcutofffrequency(阻帶截止頻率);δsisthestopbandtolerance,thatisTransitionbandThefrequencyrangeofωpc

≤ω≤ωsciscalledthetransitionband;6LowpassfilterspecificationMaximumpassbandattenuationαpMinimumstopbandattenuationαpCommonly,themaximummagnituderesponseisassumedtobenormalizedtounity.Foralowpassfilter,wehave7DesignStagesforDigitalFilterDesignstagesAnalogfilterapproximations,includingButterworth,Chebyshevandelliptic.Continuous-timetodiscrete-timetransformation,includingimpulseinvarianceandbilineartransformation.Frequencytransformations,thatis,transformingalowpassfilterintoahighpassorbandpassorbandstopfilter.81AnalogButterworth(巴特沃思)LowpassFiltersTheButterworthlowpassfilterhasseveralproperties:Allpoles,andnozero.Noripples(波紋)inthepassbandandstopband.TheButterworthlowpassfilterisdefinedinamagnitude-squaredfunction(幅度平方函數(shù)): whereNisa

positiveintegerandiscalledtheorderofthefilter,Ωcisthepassbandcutofffrequency91AnalogButterworthLowpassFiltersBecauseThecutofffrequencyΩciscalledthehalf-powerfrequencypointofthefilter.ThemaximumpassbandattenuationisthefrequencyΩcisalsocalled3-dBcutofffrequencyor3-dBbandwidthofthefilter.101AnalogButterworthLowpassFiltersInpracticalapplications,theanaloglowpassfilterisspecifiedbythespecificationsasfollows:Ωpc:thepassbandcutofffrequencyinrad/s;αp:themaximumpassbandattenuationindB;Ωsc:thestopbandcutofffrequencyinrad/s;αs:theminimumstopbandattenuationindB;TousetheButterworthlowpassfiltertoapproximatealowpassfilter,weshouldobtaintheorderNandthe3-dBcutofffrequencyΩc.111AnalogButterworthLowpassFiltersThemagnituderesponseofthefilterisThemaximumpassbandattenuationαparrivesatΩ=Ωpc,whichisrepresentedasTheminimumstopbandattenuationαsarrivesatΩ=Ωsc,whichisrepresentedas(7.141)(7.142)(7.140)121AnalogButterworthLowpassFiltersSolvingtheabovetwoequations,wehave131AnalogButterworthLowpassFiltersTherearetwochoicestodetermineΩcSubstitutingNinEq.(7.141),wehaveUsingthisformulatodetermineΩc,αpisexactlymetatΩ=Ωpcandαsisexceededforthestopband,providedthatαp≤3dB.SubstitutingNinEq.(7.142),wehaveUsingthisformulatodetermineΩc,αsisexactlymetatΩ=Ωscandαpisexceededforthepassband,providedthatαs≤3dB.141AnalogButterworthLowpassFiltersTodeterminethetransferfunctionHa(s)ofthefilter,wesubstitutes=jΩ,thereforeThepolesofthemagnitude-squaredfunctionHa(s)Ha(-s)aregivenbyFigure3showsallthepoleswhenN=3.15Figure3Poleplotforathird-orderButterworthfilterk=0k=1k=2k=3k=4k=5σjΩ161AnalogButterworthLowpassFiltersThe2Npolesequallyspacedinangleonacircleinthes-plane.Theyaresymmetricabouttheimaginaryaxis.Inordertoobtainastablesystem,wechoosethenpolesontheleft-sideofthes-plane.Then,wegetthetransferfunctionasfollowswhere171AnalogButterworthLowpassFiltersTheattenuationoftheButterworthapproximationincreasesmonotonically(單調(diào)地)withfrequency.Anditincreasesveryslowlyinthepassbandandquicklyinthestopband.Ifonewantstoincreasetheattenuationonehastoincreasethefilterorder.The3-dBbandwidthisunrelatedtothefilterorder.18Figure4Themagnitude-frequencyresponseofButterworthLowpassFilters

192AnalogChebyshev(切貝雪夫)LowpassFiltersTheChebyshev-Ilowpassfiltershaveequiripple(等波紋旳)magnituderesponseinthepassbandandmonotonic(單調(diào)旳)magnituderesponseinthestopband.Themagnitude-squaredresponseofananalogChebyshev-IlowpassfilterisgivenbywhereNistheorderofthefilter,εisapositiveandlessthanunitynumberwhichisthepassbandripplefactor,Ωcisthepassbandcutofffrequencyatwhichtheattenuationofthemagnituderesponseisnotnecessarytobe3-dB.ThefunctionTN(x)isanNth-orderChebyshevpolynomialdefinedby(7.147)202AnalogChebyshevLowpassFiltersThecharacteristicsoftheChebyshev-Ifiltersareasfollows:AtΩ=0,|Ha(j0)|=1forNoddand|Ha(j0)|=forNeven;AtΩ=Ωc,|Ha(jΩc)|=forallN;Withinthepassbandof0≤Ω≤Ωc,|Ha(jΩ)|oscillatesbetween1and;ForΩ>Ωc,|Ha(jΩ)|approacheszeromonotonicallyandrapidly;AtΩ=Ωs(thestopbandcutofffrequency),|Ha(jΩs)|=1/A.21Figure5AnalogChebyshevLowpassFilters222AnalogChebyshevLowpassFiltersInthedesignofChebyshev-Ilowpassfilter,thespecificationsaregivenby:Ωc:thepassbandcutofffrequencyinrad/s;αp:thepassbandrippleindB;Ωs:thestopbandcutofffrequencyinrad/s;αs:theminimumstopbandattenuationindB;Todesignthefilter,theorderNandripplefactorεshouldbedetermined.232AnalogChebyshevLowpassFiltersSinceThenAndTherefore242AnalogChebyshevLowpassFiltersToobtainthetransferfunctionHa(s)oftheChebyshev-Ifilter,wesubstituteΩ=s/jintoEq.(7.147)(p19),andthenwegetThereare2Npolesofthemagnitude-squaredfunctionHa(s)Ha(-s),Theyarespacedonaellipseinthes-planeandsymmetricabouttheimaginaryaxis.Inordertoobtainastablesystem,wechoosetheNpolesontheleft-sideofthes-planeandgetthetransferfunctionasfollows.252AnalogChebyshevLowpassFilterswhereand

26兩種經(jīng)典模擬濾波器兩種經(jīng)典模擬濾波器:Butterworth(巴特沃思)濾波器:幅頻特征單調(diào)下降,但衰減特征較差;Chebyshev(切貝雪夫)濾波器:在通帶(或阻帶)中幅頻特征單調(diào)下降,在阻帶(或通帶)中有波紋,衰減特征好于巴特沃思濾波器;27Analog-to-DigitalFilterTransformationsThecontinuous-timetodiscrete-timetransformationsincludethreesteps:Transformationsofspecificationsindiscrete-timedomainintoonesincontinuous-timedomain.Designingtheanalogfilteraccordingtothespecificationsincontinuous-timedomain.Transformthefilterinsdomainintotheoneinzdomain.Themainmethodsoftransformationshavetwokinds:Impulse-invariancemethodBilineartransformationmethod28ztransformandLaplacetransformLaplacetransformztransformTheLaplacetransformofx(nT)isThereforetherelationshipbetweentheLaplacetransformandztransformofx(nT)is29ztransformandLaplacetransformLetandTherelationshipbetweenrandσIfσ=0(s平面旳虛軸),r=1(z平面單位圓上);Ifσ>0(s平面旳右半平面),r>1(z平面單位圓外);Ifσ<0(s平面旳左半平面),r<1(z平面單位圓內(nèi)).TherelationshipbetweenωandΩ:ω=ΩT

IfΩ=0(s平面旳實(shí)軸),ω=0(z平面正實(shí)軸)

Ω=Ω0(s平面平行于實(shí)軸旳直線(xiàn)),ω=Ω0T(z平面始于原點(diǎn)角度為ω=Ω0T旳輻射線(xiàn)).30ztransformandLaplacetransformΩ

從–π/T

增長(zhǎng)到π/T,相應(yīng)旳,ω

從–π增長(zhǎng)到π,即s平面寬為2π/T旳一種水平條帶相當(dāng)于z平面輻角轉(zhuǎn)了一周,即整個(gè)z平面。所以Ω

每增長(zhǎng)一種抽樣角頻率Ωs=2π/T,則

ω

增長(zhǎng)2π。所以從s平面到z平面旳映射是多值映射。317.1.1FilterDesignbyImpulseInvarianceThetransferfunctionoftheanalogfiltercanbeexpressedintermsofapartial-fractionexpansionasfollowsThecorrespondingimpulseresponseis327.1.1FilterDesignbyImpulseInvarianceSamplingtheanalogimpulseresponse,wecanobtainthediscrete-timeimpulseresponseThecorrespondingdiscrete-timetransferfunctionis33s平面旳單極點(diǎn)s=pl變換為z平面上z=eplT處旳單極點(diǎn);Ha(s)與H(z)旳部分分式旳系數(shù)是相同旳;假如模擬濾波器是穩(wěn)定旳,即全部極點(diǎn)s=pl旳實(shí)部不大于零,則全部z=eplT均在單位圓內(nèi),即變換后旳數(shù)字濾波器也是穩(wěn)定旳。7.1.1FilterDesignbyImpulseInvariance347.1.1FilterDesignbyImpulseInvarianceInordertoobtainthesamepassbandgainforthecontinuous-anddiscrete-timefilters,weshouldusethefollowingexpressionforHd(z):35Example Transformthecontinuous-timelowpassfiltertransferfunctiongivenby intoadiscrete-timetransferfunctionusingtheimpulseinvariancetransformationmethodwithΩs=10rad/s.Solution

36Example(cont.)37Example(cont.)387.1.2BilinearTransformationThebilineartransformationmethodavoidstheproblemofaliasing.Inthebilineartransformationmethod,theentires-planeismappedintotheentirez-plane.Thelefthalfs-planemapsintotheinterioroftheunitcircleinthez-plane;Therighthalfs-planemapsintotheexterioroftheunitcircleinthez-plane;Theimaginaryaxisofthes-planemapsontotheunitcircle;397.1.2BilinearTransformationThebilineartransformationisdefinedasIfthecontinuous-timetransferfunctionisHa(s),thenFrequencytransformationrelation407.1.2BilinearTransformationThatisSinceWeshouldchoose417.1.2BilinearTransformationInconclusion,thebilineartransformationofacontinuous-timetransferfunctionintoadiscrete-timetransferfunctionisThereforethebilineartransformationmapsanalogfrequenciesintodigitalfrequenciesasfollows:Forhighfrequencies,thisrelationshipishighlynonlinear.Thebilineartransformationmethodavoidstheproblemofaliasing,butthepricepaidforthisistheintroductionofadistortioninfrequencyaxis,knownasthewarping.427.1.2BilinearTransformationThewarpingeffectcanbecompensatedbyprewarpingthefrequencyspecifications.ThestepsarePrewarpthepassbandandstopbandfrequenciesandobtainΩapand

Ωarthroughthefollowingmapping:GenerateHa(s)satisfyingthespecificationsforthefrequenciesΩapandΩar;ObtainHd(z)byreplacingswithinHa(s). 437.2DesignofFIRFiltersbyWindowingBasicconception Inordertodesignthefrequencyresponsessatisfyingtheprescribedspecifications,thefilterorderandmultipliercoefficientsshouldbedetermined.CharacteristicsofFIRfiltersItispossibletoobtainexactlinearphase.FIRsystemsarealwaysstable.FastalgorithmssuchasFFTcanbeused.Ahigherordermeansmoredelays,multipliersandadders.Designapproaches frequencysampling,windowfunctions,maximallyflatapproximation.44IdealcharacteristicsofstandardfiltersFourcommonlyusedFIRfiltersLowpassfiltersHighpassfiltersBandpassfiltersBandstopfiltersFouriertransformpair45Fouridealfilters—lowpassFrequencyresponseImpulseresponse146Figure--impulseresponseoflowpassfilter47Fourstandardfilters—highpassFrequencyresponseImpulseresponse148Figure--impulseresponseofhighpassfilter49Fourstandardfilters—bandpassFrequencyresponseImpulseresponse150Figure--impulseresponseofbandpassfilter51Fourstandardfilters—bandstopFrequencyresponseImpulseresponse152Figure--impulseresponseofbandstopfilter53NotesThedurationsoftheimpulseresponsesofthefourkindsoffiltersareinfinite.Theimpulseresponsesarenoncausal.So,allthesefourkindsoffiltersareidealonesandcannotberealized.54PropertiesofLinearPhaseFIRFiltersThelinear-phaseFIRfilters:Thisequationshowsthattheh(n)ofalinear-phaseFIRfilterissymmetricorantisymmetricaboutM/2.Therearefourkindsoflinear-phaseFIRfilters,whichisshowninthefigures.55PropertiesofLinearPhaseFIRFiltersnh(n)210345678nh(n)2103456789nh(n)210345678nh(n)2103456789symmetricantisymmetricMevenModdTypeITypeIITypeIIITypeIV56Frequencysampling(頻率采樣法)工程上,常給定頻域上旳技術(shù)指標(biāo),所以采用頻域設(shè)計(jì)更直接。基本思想:使所設(shè)計(jì)旳FIR數(shù)字濾波器旳頻率特征在某些離散頻率點(diǎn)上旳值精確地等于所需濾波器在這些頻率點(diǎn)處旳值,在其他頻率處旳特征則有很好旳逼近。Step1Step2Step3Step4采樣IDFTFT57FrequencysamplingLetHd(ω)bethedesiredfrequencyresponse.ThedesignapproachoffrequencysamplingisjusttosampletheHd(ω).SupposethatH(k)aresamplesoftheHd(ω),i.e.LetthenA(k)=|H(k)|

θ(k)=arg[H(k)]

A(k)isthemagnitudeoftheH(k)

andθ(k)itsphase.58FrequencysamplingWecangettheimpulseresponseh(n)fromH(k)usingIDFT

ThentheztransformofthedesignedFIRfilter

wecanalsogetthefrequencyresponseH(ejω)

59ForexampleRipplesinthepassbandandstopband.60FrequencysamplingIfthelinearphaseisrequired,A(k)andθ(k)mustsatisfytheconditionsforlinearphase.Fourtypesoflinear-phasefiltersTypeI:theorderMisevenandtheh(n)issymmetric.TypeII:theorderMisoddandtheh(n)issymmetric.TypeIII:theorderMisevenandtheh(n)isantysymmetric.TypeIV:theorderMisoddandtheh(n)isantysymmetric.Relationbetweenthelengthofh(n),N,andorderM

N=M+161Frequencysampling—typeIInthiscase,theorderMisevenandtheh(n)issymmetric.ReviewThefrequencyresponseH(ω)issymmetricaboutω=0andω=π.62Frequencysampling—typeIPhaseMagnitude63Frequencysampling—typeIIInthiscase,theorderMisoddandtheh(n)issymmetric.ReviewThefrequencyresponseH(ω)issymmetricaboutω

=

0andantisymmetricaboutω=π.So,H(ω)=0,atω=π.64Frequencysampling—typeIIPhaseMagnitudeBecauseH(ω)=0,atω=π,highpassandbandstopfilterscannotberealizedinTypeIIfilters.65Frequencysampling—typeIIIInthiscase,theorderMisevenandtheh(n)isantisymmetric.ReviewThefrequencyresponseH(ω)isantisymmetricaboutω

=

0andω=π.So,H(ω)=0,atω

=

0andω=π.66Frequencysampling—typeIIIPhaseMagnitudeBecauseH(ω)=0,atω=0andω=π,lowpass,highpassandbandstopfilterscannotberealizedinTypeIIIfilters.67Frequencysampling—typeIVInthiscase,theorderMisoddandtheh(n)isantisymmetric.ReviewThefrequencyresponseH(ω)isantisymmetricaboutω

=

0andsymmetricatω=π.So,H(ω)=0,atω

=

0.68Frequencysampling—typeIVPhaseMagnitudeBecauseH(ω)=0,atω=0,lowpassandbandstopfilterscannotberealizedinTypeIVfilters.

6970h(n)MH(ω)旳對(duì)稱(chēng)性可實(shí)現(xiàn)旳濾波器I型偶對(duì)稱(chēng)偶數(shù)在ω=0及ω=π處偶對(duì)稱(chēng)四種濾波器都可設(shè)計(jì)II型偶對(duì)稱(chēng)奇數(shù)在ω=0處偶對(duì)稱(chēng),在ω=π處奇對(duì)稱(chēng)不能設(shè)計(jì)高通和帶阻,可設(shè)計(jì)低、帶通濾波器III型奇對(duì)稱(chēng)偶數(shù)在ω=0及ω=π處奇對(duì)稱(chēng)不能設(shè)計(jì)低通、高通和帶阻,只能設(shè)計(jì)帶通濾波器IV型奇對(duì)稱(chēng)奇數(shù)在ω=0處奇對(duì)稱(chēng),在ω=π處偶對(duì)稱(chēng)不能設(shè)計(jì)低通和帶阻,可設(shè)計(jì)高通、帶通濾波器DesignofLinear-PhaseFIRfiltersusingwindowsAllidealfiltershaveinfiniteduration,sotheycannotberealized.Truncatingtheimpulseresponseh(n),wecangetitsapproximationwithfiniteduration. whereMisthefilterorderandassumingthatitiseven.Thetransferfunctionis72DesignofLinear-PhaseFIRfiltersusingwindowsAfterh(n)istruncated,thesystemisstillnoncausal.Inordertomakeitcausal,wecanshiftitrightbyM/2,withouteitherdistortingthefiltermagnituderesponseordestroyingthelinear-phaseproperty.Thereareripplesclosetothebandedgesinthemagnituderesponse.TheseripplesarereferredtoasGibbs’oscillations.AndtheamplitudesofGibbs’oscillationsdonotdecreaseeventhefilterorderMisincreased.73DesignofLinear-PhaseFIRfiltersusingwindowsMultiplyingtheimpulseresponseh(n)byawindowfunctionw(n),thatis, wecanimprovethemagnituderesponse.ThemultiplicationinthetimedomaincorrespondstoperiodicconvolutionintegralofH(ejω)andW(ejω)inthefrequencydomain,thatis74Magnituderesponsesofawindowfunction|W(ejω)|ωMainlobeSidelobe(主瓣)(旁瓣)Magnituderesponsesofawindowfunction75正肩峰負(fù)肩峰76DesignofLinear-PhaseFIRfiltersusingwindows最大旁瓣旳相對(duì)幅度越小(即能量越集中在主瓣上),起伏振蕩旳幅度越小,阻帶衰減越多;窗函數(shù)頻譜旳主瓣越窄,過(guò)渡帶越陡;這兩者是矛盾旳。|W(ejω)|ωMainlobeSidelobe(主瓣)(旁瓣)771Rectangularwindow(矩形窗)TherectangularwindowfunctionThefrequencyresponseoftherectangularwindow781Rectangularwindow792TriangularwindowandBartlettwindowThetriangularwindowfunctionTheBartlettwindowfunction802Triangularwindow(三角形窗)81Bartlettwindow823HammingandHanningwindowsThewindowfunction

with0≤α≤1.Whenα=0.54,itiscalledtheHammingwindow(海明窗).Whenα=0.5,itiscalledtheHanningwindow(漢寧窗).83Hammingwindows(海明窗)843HammingandHanningwindowsThetransitionband(過(guò)渡帶)oftheHammingwindowislargerthanthatoftherectangularwindows,duetoitswidermainlobe.TheratiobetweentheamplitudesofthemainandsecondarylobesoftheHammingwindowismuchlargerthanfortherectangularwindow.Sothestopbandattenuation(衰減)fortheHammingwindowislargerthantheattenuationfortherectangularwindow.85Hanningwindows(漢寧窗)864Blackmanwindow(布萊克

溫馨提示

  • 1. 本站所有資源如無(wú)特殊說(shuō)明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請(qǐng)下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請(qǐng)聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶(hù)所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁(yè)內(nèi)容里面會(huì)有圖紙預(yù)覽,若沒(méi)有圖紙預(yù)覽就沒(méi)有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 人人文庫(kù)網(wǎng)僅提供信息存儲(chǔ)空間,僅對(duì)用戶(hù)上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對(duì)用戶(hù)上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對(duì)任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請(qǐng)與我們聯(lián)系,我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶(hù)因使用這些下載資源對(duì)自己和他人造成任何形式的傷害或損失。

評(píng)論

0/150

提交評(píng)論