简介:Foranelectronicchartdisplayandinformationsystem(ECDIS),themostimportantfunctionistheroutedesign.Inthispaper,anovelapproachtodevelopingECDISwasgivenandthenaneffectivemethodwasproposedforroutedesignandautomaticcheck.Themathematicalmodelandlogicalrelationshipforcoordinatetransformationwerealsoprovided.TheywereadoptedinECDISdeveloping,whichwasbasedontheresearchoftheVxworksreal-timeoperatingsystem.Theactualsystemapplicationimpliesthatthemethodispracticalandcanhelpnavigationverywell.
简介:音调模型(TM)集成是为官员语音识别的一项重要任务。有效有区别地使用当把TM分数集成到多传递语音识别时,放大因素训练被证明了。而且,可伸缩的上下文依赖者(CD)能更好被申请在模型之间的插值。这条途径的一限制是很多参数将被介绍,它使这种技术容易到训练过度。在这份报纸,我们建议由使用自动地导出的语音的决定树导致上下文依赖者模型重量。在每个树节点的问题被选择在训练数据上最小化期望的识别错误。首先订客观功能被用于问题修剪使树造有效的最小的电话错误(MPE)的近似。连续官员语音识别上的试验性的结果证明方法能够导致最关键的语音的上下文并且与少得多的参数获得重要错误减小,与由使用放大参数的手工地设计的上下文依赖者获得了那相比。
简介:Baosteeldevelopedadigitalautomaticanalysistechniqueformaceralspecificationin2002.Thisanalysissystemcombinesdigitalimageprocessing,graphics,databases,expertsystems,artificialintelligenceandotheradvancedtechnologies.After6yearsofapplicationincokeproduction,thesystemproveditselfsuccessfulincoalqualitytestingandcoalblendingguidanceonmaceral.However,duringthislongprocess,someinadequacieswerefoundthatimpactedtheprecisionandaccuracyoftheanalysis.So,in2008Baosteelbegantoworkonimprovingthecoalmaceralanalysissystem.Theimprovementsincludedthefollowing:furtherupgradingandenhancingtheanalysisperformanceofmicroscopicimages;extendingthegraylevelstoincreasethereflectancemeasurementaccuracy64times;changingthefocusmethodandeffectivelyeliminatingtheinterferenceofhalo.Inaddition,animprovedimagerecognitionmethodwasadoptedtomaketheextractionofvitrinitemoreaccurateandanewmodelofcoalconstituentalgorithmwasaddedwhichcanaccuratelydeterminethecompositionofmaceral(exinite,vitrinite,inertinite).Sincetheseimprovementswerecompleted,thesystemhasachievedhigherautomation,speedandaccuracy,collectedmoreinformationandperformedmoreaccuratemaceralanalysisforcokeproduction.Meanwhile,theimprovedsystemhasprovidedareliableanalyticalbasisforthefurtherstudyontherelationshipbetweencokequalityandcoalblending.
简介:Aneuro-fuzzysystemmodelbasedonautomaticfuzzyclusteringisproposed.Ahybridmodelidentificationalgorithmisalsodevelopedtodecidethemodelstructureandmodelparameters.Thealgorithmmainlyincludesthreeparts:1)AutomaticfuzzyC-means(AFCM),whichisappliedtogeneratefuzzyrulesautomatically,andthenfixonthesizeoftheneuro-fuzzynetwork,bywhichthecomplexityofsystemdesignisreducesdgreatlyatthepriceofthefittingcapability;2)Recursiveleastsquareestimation(RLSE).ItisusedtoupdatetheparametersofTakagi-Sugenomodel,whichisemployedtodescribethebehaviorofthesystem;3)Gradientdescentalgorithmisalsoproposedforthefuzzyvaluesaccordingtothebackpropagationalgorithmofneuralnetwork.Finally,modelingthedynamicalequationofthetwo-linkmanipulatorwiththeproposedapproachisillustratedtovalidatethefeasibilityofthemethod.
简介:Currently,schemaintegrationframeworksuseapproacheslikerule-based,machinelearning,etc.Thispaperpresentsanontology-basedwrapper-mediatorframeworkthatusesboththerule-basedandmachinelearningstrategiesatthesametime.Theproposedframeworkusesglobalandlocalontologiesforresolvingsyntacticandsemanticheterogeneity,andXMLforinteroperability.Theconceptsinthecandidateschemasaremergedonthebasisofthesimilaritycoefficient,whichiscalculatedusingthedefinedrulesandthepriormappingsstoredinthecase-base.
简介:基于各种各样的算法的全面研究,传统的眼睛的光测量仪器的自动识别被认识到。拿一台通用工具显微镜(UTM)是的透镜看法图象一个例子,2层为数据读的自动识别模型在采用一系列预处理算法以后被建立。这个模型是匹配方法和并发的背繁殖(BP)的基于关联的模板的最佳的联合神经网络。多重互补特征抽取在产生并发的网络的特徵向量被使用。以便改进容错能力,旋转不变的特征从位字符和轮廓特征的一个4-dimensional组基于Zernike时刻被提取也被获得。而且,操作时间和读物精确性能被设置阀值价值动态地调整。试验性的结果显示最新发达的算法有最佳的识别精确和工作速度。平均读物比率能完成97.23%。没有修改他们的原来的结构,识别方法能自动地很快并且稳定地获得光测量仪器的结果,它满足应用程序要求。
简介:Uptonow,detailedstrategiesandalgorithmsofautomaticchangedetectionforroadnetworksbasedonGIShavenotbeendiscussed.Thispaperdiscussestwodifferentstrategiesofautomaticchangedetec-tionforimageswithlowresolutionandhighresolutionusingoldGISdata,andpresentsabufferdetectionandtracingalgorithmfordetectingroadfromlow-resolutionimagesandanewprofiletracingalgorithmfordetectingroadfromhigh-resolutionimages.Forfeature-levelchangedetection(FL-CD),aso-calledbufferdetectionalgo-rithmisproposedtodetectchangesoffeatures.SomeideasandalgorithmsofusingGISpriorinformationandsomecontextinformationsuchassubstructuresofroadinhigh-resolutionimagestoassistroaddetectionandextractionaredescribedindetail.
简介:在这份报纸,混合自动优化策略为设计被建议在水下机器人线。Isight作为一个集成平台被介绍。这个平台的构造包括UG6.0,GAMBIT2.4.6和FLUENT12.0基于用户编程和几个商业软件。一个聪明的参数优化方法,粒子群优化,被合并到平台。为了验证策略,求婚了,模拟被进行在上在水下机器人为5470建模,它从DTRCSUBOFF工程发源。与自动优化平台,最小的抵抗作为优化目标被拿;是的湿表面区域限制条件;象设计变量的船身前半部,最大的身体半径和船体后部最小半径的长度。随CFD计算,RANS方程和标准骚乱模型被用于直接数字的模拟。由模拟结果的分析,平台效率高,这被结束。通过平台,为线的设计的许多计划被产生,最佳的解决方案被完成。聪明的优化算法和数字模拟的联合保证一个全球最佳的答案并且改进寻找的答案的效率。
简介:
简介:
简介:Sinceitsintroductionin1967,intra-aorticballoonpumping(IABP)hasbecomethemainstayinthemanagementofacuteleftventriculardysfunctioninthecriticalcaresetting.Typically,presentIABPsystemsareunabletoprovideconsistent,effectivecar-diacassistanceinthepresenceofarrhythmiasorrapidlychangingheartrates.Anew-generation,closedloop,fullyautomaticintra-aorticballoonpump(CL-IABP)
简介:Opticalmotioncaptureisanincreasinglypopularanimationtechnique.Inthelastfewyears,plentyofmethodshavebeenproposedforkey-frameextractionofmotioncapturedata,anditisacommonmethodtoextractkey-frameusingquaternion.Here,onemaindifficultyisduetothefactthatpreviousalgorithmsoftenneedtomanuallysetvariousparameters.Inaddition,itisproblematictopredefinetheappropriatethresholdwithoutknowingthedatacontent.Inthispaper,wepresentanoveladaptivethreshold-basedextractionmethod.Key-framecanbefoundaccordingtoquaterniondistance.Weproposeasimpleandefficientalgorithmtoextractkey-framefromamotionsequencebasedonadaptivethreshold.Itisconvenientwithnoneedtopredefineparameterstomeetcertaincompressionratio.Experimentalresultsofmanymotioncaptureswithdifferenttraitsdemonstrategoodperformanceoftheproposedalgorithm.Ourexperimentsshowthatonecantypicallycutdowntheprocessofextractionfromseveralminutestoacoupleofseconds.
简介:Identifyingeachprocessandtheirconstraintrelationsfromthecomplexwiringharnessdrawingsquicklyandaccuratelyisthebasisforformulatingprocessroutes.Accordingtotheknowledgeofautomotivewiringharnessandthecharacteristicsofwiringharnesscomponents,weestablishedthemodelofwiringharnessgraph.Thenweresearchthealgorithmofidentifyingtechnologyprocessesautomatically,finallywedescribetherelationshipsbetweenprocessesbyintroducingtheconstraintmatrix,whichisinordertolayagoodfoundationforharnessprocessplanningandproductionscheduling.
简介:AUTOMATICSELECTIONOFPARAMETERSINSPLINEREGRESSIONVIAKULLBACK-LEIBLERINFORMATIONAUTOMATICSELECTIONOFPARAMETERSINSPLINEREGRESSIO...