簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--自動(dòng)焊接操作系統(tǒng)精彩內(nèi)容。

簡(jiǎn)介：字?jǐn)?shù)英文字?jǐn)?shù)英文23222322單詞，單詞，1306013060字符；中文字符；中文39583958漢字漢字出處出處JACEKJACEKOSKARBSKIAUTOMATICOSKARBSKIAUTOMATICROADROADTRAFFICTRAFFICSAFETYSAFETYMANAGEMENTMANAGEMENTSYSTEMSYSTEMININURBANURBANAREASJMATECAREASJMATECWEBWEBOFOFCONFERENCES2017,1220300716CONFERENCES2017,1220300716外文文獻(xiàn)外文文獻(xiàn)AUTOMATICROADTRAFFICSAFETYMANAGEMENTSYSTEMINURBANAREASABSTRACTTRAFFICINCIDENTSANDACCIDENTSCONTRIBUTETODECREASINGLEVELSOFTRANSPORTSYSTEMRELIABILITYANDSAFETYTRAFFICMANAGEMENTANDEMERGENCYSYSTEMSONTHEROAD,USING,AMONGOTHERS,AUTOMATICDETECTION,VIDEOSURVEILLANCE,COMMUNICATIONTECHNOLOGIESANDINSTITUTIONALSOLUTIONSIMPROVETHEORGANIZATIONOFTHEWORKOFVARIOUSDEPARTMENTSINVOLVEDINTRAFFICANDSAFETYMANAGEMENTAUTOMATIONOFINCIDENTMANAGEMENTHELPSTOREDUCETHETIMEOFARESCUEOPERATIONASWELLASOFTHENORMALIZATIONOFTHEFLOWOFTRAFFICAFTERCOMPLETIONOFARESCUEOPERATION,WHICHALSOAFFECTSTHEREDUCTIONOFTHERISKOFSECONDARYACCIDENTSANDCONTRIBUTESTOREDUCINGTHEIRSEVERITYTHEPAPERPRESENTSTHEPOSSIBILITYOFINCLUDINGCITYTRAFFICDEPARTMENTSINTHEPROCESSOFINCIDENTMANAGEMENTTHERESULTSOFRESEARCHONTHEAUTOMATICINCIDENTDETECTIONINCITIESAREALSOPRESENTED1INTRODUCTIONTRANSPORTSYSTEMSAREAKEYELEMENTINMEETINGBASICSOCIALNEEDSRELATEDTOPOPULATIONMOBILITYANDSUPPLIESTHEYFACILITATETHEDEVELOPMENTOFECONOMICACTIVITYANDCONTRIBUTETOUPGRADINGTHEQUALITYOFLIFESOCIETYEXPECTSAHIGHLEVELOFRELIABILITYOFTRANSPORTWHILETRAVELINGTOWORK,SCHOOLSANDFORRECREATIONTODAY,SYSTEM,WECANIDENTIFYMANYINTERRELATEDANDMUTUALLYINTERACTINGELEMENTSSTRUCTURESOFORGANIZATIONOFTRAFFICSAFETYMANAGEMENTSERVICES,ASWELLASLEGALBASESANDPROCEDURESTHATDETERMINETHEORGANIZATION,COMPETENCESANDCOLLABORATIONOFSUCHSERVICES,TOOLSINFLUENCINGTHEIMPROVEMENTOFTRAFFICSAFETYMANAGEMENTATTHESTRATEGICLEVELEGREGIONAL,LOCALANDSECTORALPROGRAMSOFTRAFFICSAFETYMANAGEMENT,INCLUDINGMONITORINGOFTHEIRIMPLEMENTATIONANDTHEOPERATIONALLEVELCURRENTMONITORINGANDSUPERVISION,BROADERSUPPORTTOOLSFORSYSTEMMANAGEMENTEGDATABASES,INFORMATIONSYSTEMS,EXPERTSYSTEMS,GUIDELINESANDEXAMPLESOFGOODPRACTICE,ITSSERVICESBASEDONSCIENTIFICRESEARCH,MANAGEMENTMETHODSFORTHEWHOLESYSTEMANDITSSELECTEDELEMENTSINCLUDINGRISKMANAGEMENTTHENATIONALROADSAFETYPROGRAMFOR20132020INDICATEDTHATTHEPROCESSOFIMPROVINGTRAFFICSAFETYREQUIRESTHEUSEOFUPTOTHREECONSECUTIVEANDINTERLINKEDCOMPONENTSINSTITUTIONALMANAGEMENTFUNCTIONS,SPECIFICACTIONSINTERVENTIONSANDOUTCOMESBASICFUNCTIONSOFINSTITUTIONALMANAGEMENTINCLUDEDCOORDINATION,LEGISLATION,FINANCINGANDPROVISIONOFRESOURCES,PROMOTIONANDCOMMUNICATION,MONITORINGANDEVALUATION,ASWELLASRESEARCH,DEVELOPMENTANDKNOWLEDGETRANSFERTHEDIAGNOSISOFTHEEXISTINGSYSTEMOFSAFETYMANAGEMENTCARRIEDOUTINTHEFRAMEWORKOFTHENATIONALROADSAFETYPROGRAMHASSHOWNTHATEACHOFTHESEFUNCTIONSNEEDSTOBEIMPROVEDINTERMSOFIMPLEMENTATIONOFMEASURESINCLUDEDINTHEPROGRAMTHESEFUNCTIONSAREPERFORMEDINDIFFERENTPROPORTIONSDEPENDINGONTHEPARTICULARINSTITUTIONANDTHELEVELOFPUBLICADMINISTRATIONTHEYHIGHLIGHTEDTHENEEDTOIMPROVEORGANIZATIONALSTRUCTURESOFTRAFFICSAFETYANDCOORDINATIONWITHTHENATIONALROADSAFETYCOUNCILASALEADINGINSTITUTION,ASWELLASWITHREGIONALROADSAFETYCOUNCILSASTHEREALLEADERSOFTHEREGION–WITHTHESUPPORTOFRESEARCH

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--注塑模具自動(dòng)裝配造型精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--掘進(jìn)機(jī)自動(dòng)切割控制組精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--基于神經(jīng)網(wǎng)絡(luò)和遺傳算法的模糊系統(tǒng)的自動(dòng)設(shè)計(jì)精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--平衡懸臂施工混凝土橋撓度和溫度的自動(dòng)監(jiān)測(cè)精彩內(nèi)容。

簡(jiǎn)介：出處出處SHERMEAEANOVELMETHODFORAUTOMATICMODULATIONRECOGNITIONJAPPLIEDSOFTCOMPUTING,2012,121453461中文中文5015字一種新的自動(dòng)調(diào)制識(shí)別的方法一種新的自動(dòng)調(diào)制識(shí)別的方法摘要數(shù)字信號(hào)的自動(dòng)調(diào)制識(shí)別在各種各樣的應(yīng)用中起著重要的作用。本文研究一種用于數(shù)字調(diào)制的準(zhǔn)確的系統(tǒng)的設(shè)計(jì)。首先，它引入了一個(gè)高效的模式識(shí)別系統(tǒng)，其包括兩個(gè)主要模塊特征提取模塊和分類模塊。特征提取模塊提取一種合適的組合，即高達(dá)八階的瞬時(shí)量、高達(dá)八階的高階累積量以及數(shù)字調(diào)制的瞬時(shí)特性。這些特征的組合是第一次被運(yùn)用在這一領(lǐng)域中。在分類器模塊,兩類重要的被控制的分類器，基于分類器的MLP神經(jīng)網(wǎng)絡(luò)以及多層多級(jí)支持向量機(jī)器被研究。通過(guò)實(shí)驗(yàn)性的研究，我們?yōu)榧榷ǖ恼{(diào)制的識(shí)別選擇了最合適的分類器。然后，我們提出了一個(gè)混合啟發(fā)式識(shí)別系統(tǒng)，在這個(gè)系統(tǒng)中一個(gè)優(yōu)化模塊被添加上用來(lái)改善分類器通常的表現(xiàn)。在這個(gè)模塊中，我們提出了一個(gè)叫做蜜蜂算法的新的優(yōu)化算法。該模塊通過(guò)尋找適合其判別函數(shù)的參數(shù)的最佳值來(lái)優(yōu)化分類器的設(shè)計(jì)。仿真結(jié)果表明，所提出的混合智能技術(shù)有很高的識(shí)別精度，即使是在特征不明顯的低信噪比的情況下。1引言自動(dòng)調(diào)制識(shí)別是一種識(shí)別在接收機(jī)處接收到的信號(hào)的類型的技術(shù)。它在軍事和民用領(lǐng)域都發(fā)揮著重要的作用。例如，在軍事應(yīng)用中，它可用于電子監(jiān)視，干擾識(shí)別和干擾監(jiān)控。廣泛的民用應(yīng)用包括頻譜管理，網(wǎng)絡(luò)流量管理，信號(hào)確認(rèn)，軟件無(wú)線電，智能解調(diào)器，認(rèn)知無(wú)線電等。由于在例如軟件無(wú)線電這樣的新技術(shù)中對(duì)數(shù)字信號(hào)使用的不斷增長(zhǎng)，近期的研究一直集中在識(shí)別這些信號(hào)的類型的方面。一般地，數(shù)字信號(hào)的識(shí)別方法歸為兩大類決策理論（DT）的方法和模式識(shí)別（PR）的方法。決策理論方法使用概率論和假設(shè)檢驗(yàn)來(lái)表示出識(shí)別的問(wèn)題。決策理論方法的主要缺點(diǎn)是其計(jì)算復(fù)雜性過(guò)高，缺乏模式不匹配的魯棒性以及需要設(shè)置正確的閾值的仔細(xì)的分析。模式識(shí)別方法，不管怎樣，不需要這樣復(fù)雜的處理，它們很容易實(shí)現(xiàn)。模式識(shí)別方法可以被更進(jìn)一步地分為兩個(gè)主要的子系統(tǒng)特征提取子系統(tǒng)和分類器子系統(tǒng)。前者提取特征（例如直方圖，光譜特性，瞬時(shí)特性，第二階距和第四階矩的結(jié)合，對(duì)稱性等），并且后者決定信號(hào)的組成（例如，神經(jīng)網(wǎng)絡(luò)，K近鄰，模糊邏輯分類器等）。從現(xiàn)有的出版物來(lái)看，這件事看起來(lái)很明顯，即在設(shè)計(jì)一個(gè)用來(lái)自動(dòng)識(shí)別數(shù)字信一種新的自動(dòng)調(diào)制識(shí)別的方法（1）????????BIANSIANSTNTNAIACNAIACNAA2211???????????????????????其中ACN（I）為在時(shí)間TI/FSI1,2,,NS時(shí)的歸一化的瞬時(shí)值，F(xiàn)S是采樣速率，NS是每個(gè)信號(hào)周期的采樣點(diǎn)數(shù)AT是AN（I）的閾值，低于這個(gè)值時(shí)瞬時(shí)相位的估計(jì)對(duì)噪聲是非常敏感的。它可以被用來(lái)識(shí)別ASK2和ASK4因?yàn)閷?duì)ASK2而言，其瞬時(shí)幅度的絕對(duì)值是一個(gè)常量。212AP?非弱信號(hào)段的瞬時(shí)相位的居中非線性成分的絕對(duì)值的標(biāo)準(zhǔn)差。（2）2211NTNTAPNLINLIAIAAIACC?????????????????????????C為在中的采樣點(diǎn)數(shù)（在瞬時(shí)時(shí)間時(shí)刻），表示非弱部分????INL?SFIT???TNAIA?的點(diǎn)。，在文獻(xiàn)中定義為，在這里我們將它改??0?????INLI0??????????SNISIN101進(jìn)為。??????????STNNAIASIC?10213AF?截獲的信號(hào)的非弱段的標(biāo)準(zhǔn)中心化瞬時(shí)頻率的絕對(duì)值的標(biāo)準(zhǔn)差（3）????????2211????????????????????TNTNAIANAIANAFIFCIFC???，RS是數(shù)字序列的采樣率，C是??????????????????NIFFCSCNIFNMMIFIFRIFIF11,,在中的采樣點(diǎn)數(shù)（在瞬時(shí)時(shí)間），表示非弱部分的點(diǎn)。????IFNSFIT???TNAIA?這個(gè)特征可以區(qū)分沒(méi)有頻率信息的調(diào)制方案與FSK調(diào)制方案以及FSK2和FSK4調(diào)制方案。214MAX?截取的信號(hào)段的歸一化中心瞬時(shí)幅度的功率譜密度的最大值。它被定義為

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--異步電動(dòng)機(jī)的電氣故障診斷監(jiān)測(cè)系統(tǒng)精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--帶式輸送機(jī)自動(dòng)維護(hù)系統(tǒng)方法精彩內(nèi)容。

簡(jiǎn)介：APPLIEDSOFTCOMPUTING122012453–461CONTENTSLISTSAVAILABLEATSCIVERSESCIENCEDIRECTAPPLIEDSOFTCOMPUTINGJOURNALHOMEPAGEWWWELSEVIERCOM/LOCATE/ASOCANOVELMETHODFORAUTOMATICMODULATIONRECOGNITIONATAOLLAHEBRAHIMZADESHERMEFACULTYOFELECTRICALANDCOMPUTERENGINEERING,BABOLUNIVERSITYOFTECHNOLOGY,BABOL,IRANARTICLEINFOARTICLEHISTORYRECEIVED15SEPTEMBER2009RECEIVEDINREVISEDFORM23DECEMBER2010ACCEPTED14AUGUST2011AVAILABLEONLINE30AUGUST2011KEYWORDSMODULATIONRECOGNITIONPATTERNRECOGNITIONBEESALGORITHMHIERARCHICALSUPPORTVECTORMACHINEBASEDCLASSIFIERCOMBINATIONOFTHEHIGHERORDERMOMENTSUPTOEIGHTHANDHIGHERORDERCUMULANTSUPTOEIGHTHSPECTRALCHARACTERISTICSABSTRACTAUTOMATICRECOGNITIONOFTHEDIGITALMODULATIONPLAYSANIMPORTANTROLEINVARIOUSAPPLICATIONSTHISPAPERINVESTIGATESTHEDESIGNOFANACCURATESYSTEMFORRECOGNITIONOFDIGITALMODULATIONSFIRST,ITISINTRODUCEDANEFFICIENTPATTERNRECOGNITIONSYSTEMTHATINCLUDESTWOMAINMODULESTHEFEATUREEXTRACTIONMODULEANDTHECLASSIFIERMODULEFEATUREEXTRACTIONMODULEEXTRACTSASUITABLECOMBINATIONOFTHEHIGHERORDERMOMENTSUPTOEIGHTH,HIGHERORDERCUMULANTSUPTOEIGHTHANDINSTANTANEOUSCHARACTERISTICSOFDIGITALMODULATIONSTHESECOMBINATIONSOFTHEFEATURESAREAPPLIEDFORTHEFIRSTTIMEINTHISAREAINTHECLASSIFIERMODULE,TWOIMPORTANTCLASSESOFSUPERVISEDCLASSIFIERS,IE,MULTILAYERPERCEPTRONMLPNEURALNETWORKANDHIERARCHICALMULTICLASSSUPPORTVECTORMACHINEBASEDCLASSIFIERAREINVESTIGATEDBYEXPERIMENTALSTUDY,WECHOOSETHEBESTCLASSIFIERFORRECOGNITIONOFTHECONSIDEREDMODULATIONSTHEN,WEPROPOSEAHYBRIDHEURISTICRECOGNITIONSYSTEMTHATANOPTIMIZATIONMODULEISADDEDTOIMPROVETHEGENERALIZATIONPERFORMANCEOFTHECLASSIFIERINTHISMODULEWEHAVEUSEDANEWOPTIMIZATIONALGORITHMCALLEDBEESALGORITHMTHISMODULEOPTIMIZESTHECLASSIFIERDESIGNBYSEARCHINGFORTHEBESTVALUEOFTHEPARAMETERSTHATTUNEITSDISCRIMINANTFUNCTION,ANDUPSTREAMBYLOOKINGFORTHEBESTSUBSETOFFEATURESTHATFEEDTHECLASSIFIERSIMULATIONRESULTSSHOWTHATTHEPROPOSEDHYBRIDINTELLIGENTTECHNIQUEHASVERYHIGHRECOGNITIONACCURACYEVENATLOWLEVELSOFSNRWITHALITTLENUMBEROFTHEFEATURES?2011ELSEVIERBVALLRIGHTSRESERVED1INTRODUCTIONAUTOMATICMODULATIONRECOGNITIONISATECHNIQUETHATRECOGNIZESTHETYPEOFTHERECEIVEDSIGNALATTHERECEIVERITPLAYSANIMPORTANTROLEINMILITARYANDCIVILDOMAINSFOREXAMPLE,INMILITARYAPPLICATIONS,ITCANBEEMPLOYEDFORELECTRONICSURVEILLANCE,INTERFERENCERECOGNITIONANDMONITORINGTHEWIDERANGEOFCIVILIANAPPLICATIONSINCLUDESSPECTRUMMANAGEMENT,NETWORKTRAFFICADMINISTRATION,SIGNALCONFIRMATION,SOFTWARERADIOS,INTELLIGENTMODEMS,COGNITIVERADIO,ETCDUETOTHEINCREASINGUSAGEOFDIGITALSIGNALSINNOVELTECHNOLOGIESSUCHASSOFTWARERADIO,THERECENTRESEARCHESHAVEBEENFOCUSEDONIDENTIFYINGTHESESIGNALTYPESGENERALLY,DIGITALSIGNALTYPEIDENTIFICATIONMETHODSFALLINTOTWOMAINCATEGORIESDECISIONTHEORETICDTMETHODSANDPATTERNRECOGNITIONPRMETHODSDTMETHODSUSEPROBABILISTICANDHYPOTHESISTESTINGARGUMENTSTOFORMULATETHERECOGNITIONPROBLEM1–3THEMAJORDRAWBACKSOFDTMETHODSARETHEIRTOOHIGHCOMPUTATIONALCOMPLEXITY,LACKOFROBUSTNESSTOTHEMODELMISMATCHASWELLASCAREFULANALYSISTHATAREREQUIREDTOSETTHECORRECTTHRESHOLDVALUES4PRMETHODS,HOWEVER,DONOTNEEDSUCHCAREFULTREATMENTTHEYAREEASYTOIMPLEMENTPRMETHODSCANBEFURTHERDIVIDEDINTWOMAINSUBSYSTEMSTHEFEATUREEXTRACTIONSUBSYSTEMEMAILADDRESSABRAHAMZADEHGMAILCOMANDTHECLASSIFIERSUBSYSTEMTHEFORMEREXTRACTSTHEFEATURESEGHISTOGRAMS,SPECTRALCHARACTERISTICS,INSTANTANEOUSCHARACTERISTICS,COMBINATIONOFSECONDANDFOURTHORDERMOMENT,SYMMETRY,ETC,ANDTHELATTERDETERMINESTHEMEMBERSHIPOFSIGNALEGNEURALNETWORKS,KNEARESTNEIGHBOR,FUZZYLOGICCLASSIFIER,ETC4–19FROMTHEPUBLISHEDWORKS,ITAPPEARSCLEARTHATINTHEDESIGNOFASYSTEMFORAUTOMATICRECOGNITIONOFDIGITALSIGNALTYPEMODULATION,THEREARESOMEIMPORTANTISSUES,WHICH,IFSUITABLYADDRESSED,LEADTOTHEDEVELOPMENTOFMOREROBUSTANDEFFICIENTRECOGNIZERSONEOFTHESEISSUESISRELATEDTOTHECHOICEOFTHECLASSIFICATIONAPPROACHTOBEADOPTEDLITERATUREREVIEWSHOWSTHATDESPITEITSGREATPOTENTIAL,THEAPPLICATIONOFDIFFERENTSUPERVISEDCLASSIFIERHASNOTRECEIVEDTHEATTENTIONITDESERVESINTHEMODULATIONCLASSIFICATIONTHEREFORE,INTHISPAPERWEINVESTIGATEDTHEPERFORMANCESOFMULTILAYERPERCEPTRONNEURALNETWORKMLP20,ANDSUPPORTVECTORMACHINESVM21,22INTHISPAPER,WEHAVEUSEDTHESVMSINTHESTRUCTUREOFTHEPROPOSEDHIERARCHICALCLASSIFIERCHOOSINGTHERIGHTFEATURESETISSTILLANOTHERISSUEINTHISPAPER,ASUITABLESETOFTHEINSTANTANEOUSCHARACTERISTICS,THEHIGHERORDERMOMENTSUPTOEIGHTHANDTHEHIGHERORDERCUMULANTSUPTOEIGHTHAREPROPOSEDASTHEEFFECTIVEFEATURESTURNINGBACKTOTHEDIGITALSIGNALRECOGNITIONSYSTEMS,ITISFOUNDTHAT1FEATURESELECTIONISNOTPERFORMEDINACOMPLETELYAUTOMATICWAYAND2THESELECTIONOFTHEBESTFREEPARAMETERSOFTHEADOPTEDCLASSIFIERAREGENERALLYDONEEMPIRICALLYMODELSELECTIONISSUEANOTHERISSUETHATISADDRESSEDINTHISPAPERISOPTIMIZATIONINTHISMODULEWEHAVEUSEDANEW15684946/–SEEFRONTMATTER?2011ELSEVIERBVALLRIGHTSRESERVEDDOI101016/JASOC201108025AESHERME/APPLIEDSOFTCOMPUTING122012453–461455WHEREMISTHEMEANOFTHERANDOMVARIABLETHEDEFINITIONFORTHEITHMOMENTFORAFINITELENGTHDISCRETESIGNALISGIVENBY?IN?K1SK?MIFSK7WHERENISTHEDATALENGTHINTHISSTUDYSIGNALSAREASSUMEDTOBEZEROMEANTHUS?IN?K1SIKFSK8NEXT,THEAUTOMOMENTOFTHERANDOMVARIABLEMAYBEDEFINEDASFOLLOWSMPQESP?QS?Q9WHEREPISCALLEDTHEMOMENTORDERANDSSTANDSFORCOMPLEXCONJUGATIONOFSASSUMEAZEROMEANDISCRETEBASEDBANDSIGNALSEQUENCEOFTHEFORMSKAKJBKUSINGTHEDEFINITIONOFTHEAUTOMOMENTS,THEEXPRESSIONSFORDIFFERENTORDERSMAYBEEASILYDERIVEDFOREXAMPLEM83ES5S?3EAJB5A?JB3?M83EA5J5A4BJ210A3B2J310A2B3J45AB4J5B5A3?J3A2BJ23AB2?J3B3?M83EA8J2A7B?J22A6B2?J36A5B3J560A3B5J62A2B6?J72AB7?J8B8?M83EA82A6B2?2A2B6?B810CONSIDERASCALARZEROMEANRANDOMVARIABLESWITHCHARACTERISTICFUNCTION?FTE{EJTS}11EXPANDINGTHELOGARITHMOFTHECHARACTERISTICFUNCTIONASATAYLORSERIES,ONEOBTAINSLOG?FTK1JTKRJTRR12THECONSTANTSKRIN12ARECALLEDTHECUMULANTSOFTHEDISTRIBUTIONOFSTHESYMBOLISMFORPTHORDEROFCUMULANTISSIMILARTOTHATOFTHEPTHORDERMOMENTMORESPECIALLYCPQCUMS,,S????P?QTERMS,S?,,S?????QTERMS13FOREXAMPLEC81CUMS,S,S,S,S,S,S,S?14THENTHORDERCUMULANTISAFUNCTIONOFTHEMOMENTSOFORDERSUPTOANDINCLUDINGNMOMENTSMAYBEEXPRESSEDINTERMSOFCUMULANTSASMS1,,SN??VCUM{SJ}J∈V1CUM{SJ}J∈VQ15WHERETHESUMMATIONINDEXISOVERALLPARTITIONSVV1,,VQFORTHESETOFINDEXES1,2,,N,ANDQISTHENUMBEROFELEMENTSINAGIVENPARTITIONCUMULANTSMAYBEALSOBEDERIVEDINTERMSOFMOMENTSCUMS1,,SN??V?1Q?1Q?1E???J∈V1SJ??E???J∈VQSJ??16WHERETHESUMMATIONISBEINGPERFORMEDONALLPARTITIONSVV1,,VQFORTHESETOFINDICES1,2,,NWEHAVECOMPUTEDALLOFTHEHIGHERORDERFEATURESFORTHEDIGITALCOMMUNICATIONSIGNALSTHATARECONSIDEREDTABLE1SHOWSTHESOMEOFTHETHEORETICALVALUESOFTHEHIGHERORDERSTATISTICSFORANUMBEROFTHECONSIDEREDDIGITALSIGNALTYPESTHESEVALUESARECOMPUTEDTABLE1SOMEOFTHEHIGHERORDERFEATURESFORANUMBEROFTHECONSIDEREDDIGITALSIGNALTYPESPSK2QAM16QAM64M41100M611?132?13M84131339C61162081797C80?244?1399?115C84?24417380UNDERTHECONSTRAINTOFUNITVARIANCEINNOISEFREEANDNORMALIZEDBYTHEORETICALSIGNALPOWER,IE,THESEVALUESAREOBTAINEDASSUMINGTHESIGNALISCLEANANDOFINFINITELENGTHHOWEVER,INPRACTICESIGNALSAREUSUALLYSUBJECTTOSOMETYPEOFDISTORTION,EITHERINSIDETHETRANSMITTERORDURINGTRANSMISSION,ANDAREOFFINITELENGTHFIG1SHOWSONEOFTHEHIGHERORDERFEATURESFORANUMBEROFTHECONSIDEREDDIGITALSIGNALTYPES3CLASSIFIERINTHISPAPER,TWOIMPORTANTSUPERVISEDCLASSIFIERSHAVEUSEDFOLLOWINGPHRASESDESCRIBESBRIEFLYTHESECLASSIFIERS31MLPNEURALNETWORKANMLPNEURALNETWORKCONSISTSOFANINPUTLAYEROFSOURCENODES,ONEORMOREHIDDENLAYERSOFCOMPUTATIONNODESANDANOUTPUTLAYER21THEISSUEOFLEARNINGALGORITHMANDITSSPEEDISVERYIMPORTANTFORMLPONEOFTHEMOSTPOPULARLEARNINGALGORITHMSISBACKPROPAGATIONBPALGORITHMHOWEVERUNDERCERTAINCONDITIONS,THEBPNETWORKCLASSIFIERCANPRODUCENONROBUSTCLASSIFICATIONRESULTSANDEASILYCONVERGETOALOCALMINIMUMMOREOVERITISTIMECONSUMINGINTRAININGPHASEINRECENTYEARS,NEWLEARNINGALGORITHMSHAVEBEENPROPOSEDFORNEURALNETWORKTRAININGINTHISPAPER,THERESILIENTBACKPROPAGATIONRPROPALGORITHMISUSEDASTHELEARNINGALGORITHMOFTHEMLPNEURALNETWORK26RPROPCONSIDERSTHESIGNOFDERIVATIVESASTHEINDICATIONFORTHEDIRECTIONOFTHEWEIGHTUPDATEINDOINGSO,THESIZEOFTHEPARTIALDERIVATIVEDOESNOTINFLUENCETHEWEIGHTSTEPTHEFOLLOWINGEQUATIONSHOWSTHEADAPTATIONOFTHEUPDATEVALUESOF?IJWEIGHTCHANGESFORTHERPROPALGORITHMFORINITIALIZATION,ALL?IJARESETTOSMALLPOSITIVEVALUES?IJT??????????IJT?1IFIEIWIJT?1IEIWIJT?0????IJT?1IFIEIWIJT?1IEIWIJT?0?0??IJT?1OTHERWISE17WHERE?01,0????1??,??,0,AREKNOWNASTHEUPDATEFACTORS,WIJREPRESENTSTHEWEIGHTVALUEFROMNEURONJTONEURONI,ANDEREPRESENTSTHEERRORFUNCTIONWHENEVERTHEDERIVATIVEOFTHECORRESPONDINGWEIGHTCHANGESITSSIGN,ITIMPLIESTHATTHEPREVIOUSUPDATEVALUEISTOOLARGEANDITHASSKIPPEDAMINIMUMTHEREFORE,THEUPDATEVALUEISTHENREDUCED??ASSHOWNABOVEHOWEVER,IFTHEDERIVATIVERETAINSITSSIGN,THEUPDATEVALUEIS?INCREASEDTHISWILLHELPTOACCELERATECONVERGENCEINSHALLOWAREASTOAVOIDOVERACCELERATION,INTHEEPOCHFOLLOWINGTHEAPPLICATIONOF?,THENEWUPDATEVALUEISNEITHERINCREASEDNORDECREASED?0FROMTHEPREVIOUSONENOTETHATVALUESOF?IJREMAINNONNEGATIVEINEVERYEPOCHTHISUPDATEVALUEADAPTATIONPROCESSISTHENFOLLOWEDBYTHE

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--基于單片機(jī)的火災(zāi)自動(dòng)報(bào)警系統(tǒng)精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--基于制造特征的三軸高速銑削數(shù)控自動(dòng)編程技術(shù)（有word版）精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--地面車輛自動(dòng)變速器的換擋節(jié)能研究精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--在各種嚴(yán)重路況下frid電動(dòng)汽車的電氣和機(jī)械聯(lián)合制動(dòng)精彩內(nèi)容。

簡(jiǎn)介：點(diǎn)擊查看更多外文翻譯--基于制造特征的三軸高速銑削數(shù)控自動(dòng)編程技術(shù)精彩內(nèi)容。