-
EKF UKF PF的算法仿真比较
EKF UKF PF三个算法在f(x) = 0.5 *x + 25*x/1+x^2 + 8 * cos(1.2*(k-1))下仿真比较，自己写的代码，直接可以产生很好的效果图，直接可以给导师看。
- 2020-07-04下载
- 积分：1
-
PML吸收边界三维FDTD算法MATLAB代码
PML吸收边界三维FDTD算法MATLAB代码
- 2020-06-02下载
- 积分：1
-
基于STM32的双向DC-DC变换器的设计与实现
本系统主要由 BUCK 降压模块、BOOST 升压模块、测控模块、辅助电源模块组成。其中BUCK 降压模块和BOOST 升压模块的驱动选用具有波形互补的可编程芯片IR2104、电流采样选用TI 公司专用高边电流采样芯片INA282；测控模块采用低功耗单片机STM32 对输出电压、输出电流实现闭环PI 控制。系统可以实现：在充电模式下，充电电流在 1～2A范围内步进可调且步进值为 0.05A，电流控制精度 1.30%左右；充电电流变换率为 0.87%；充电效率可达到 97.11%，具有测量、显示充电电流以及过充保护功能。在放电模式下，放电效率可达到96.54%且电压能保持在 30V目录第一章绪论1.1课题背景·*······*···*·····*···‘1.2双向DC-D变换器的研究意义1121.3国内外研究和应用现状1.4论文主要的研究内容.第二章双向DG-DG变换器拓扑结构的硏究.34662.1双向DC-DG变换器的基本原理与类型2.2双向DC-DG变换器的电路拓扑2.3双向DCDC变换器方案的设计10第三章双向DC-DC变换器硬件电路分析及参数设计.3.1双向DG-DG变换器的硬件电路分析.…123.2BUCK-B00sT电路器件的选择及参数设计3.3电流采样电路分析及参数设计173.4 MOSFET管驱动电路设计183.5辅助电源设计.19第四章双向DG-DG变换器的软件设计4.1软件设计方法214.2主函数程序设计4.3按键模式的识别.224.4恒流恒压模式的设计……第五章双向DG-DG变换器调试、实验结果与分析255.1测试仪器∴255.2测试方法255.3测试实验数据5.4测试结果分析…27第六章总结与展望6.1总结286.2展望.28[参考文献]附录(一):项目课题获奖情况及总体实物图….31附录1.1项目课题获奖情况31附录1.2双向D-DC变换器的总体实物图,34附录(二)程序清单…..35第一章绪论1.1课题背景航天器由若下分系统组成,分为有效载荷和航天器平台两大类。有效载荷主要是直接执行特殊的航天任务,而航天器平台主要由航天器结构和服务与支持系统构成。服务与支持系统主要包括电源裝置、姿态控制裝置、轨道控制装置、无线电测控装置、数据保管等等。因此,电源分系统是极其重要的,它是航大器所有能源供给装置。若电源部分工作不止常,则整体就将失去作用,变为毫无用处,电源重量占航天器重量的15%~25%。分为化学电源、太阳电池电源和核电源三类。日前世界上90%以上的航天器都采用太阳能电池阵构成的光伏电源发电系统。主功率供电回路的额定电压(母线电压)三个等级:(1)低压—28V,适用功率等级:1200W(2)中压——42或50V,适用功率等级:200水平(3)高压—100V或以上,适用功率等级:4000V水平。载人飞船氿道运行高度为300~400Km,轨道周期约为9lmin,其中轨道最长,阴影吋间37min,最短光照时间54min。飞船屯源分系统组成部分如表1所表1飞船电源分系统组成电源名称电源类型配置舱段用途备注太阳电池阵-镉镍待发段、发射段、自主主电源推进舱蓄电池系统运行段向整船供电有留轨仁务需要时,飞留轨电源太阳电池镉都轨道舱留轨使用期间船配置留轨电源,否电池系统不配置返回/着陆返回、着陆、等待期旧锌银蓄电池组返回舱电源供电补充峰值功率、应急飞应急电源锌银蓄电池组推进舱行供电目前,我国的航天电源部分调节器主要依赖于从欧洲等国家进口,需要耗费巨资,对我国载人航天的航天器产生极其不利的影响。因此,具有自主知识产权的电源部分调节器的研制,具有很重要的意义和深远的影响1.2双向DDG变换器的研究意义在传统的太阳能电池阵构成的光伏电源发电系统,传统的蓄电池充、放电模块很难保证太阳能阵在太阳光线充足时产生多余的能量不会导致航天器的过热以及储能装置蓄电池组的过允电,而且功率密度点较大,成木高,系统结构相对复杂。太阳能光伏电源发电系统是将太阳能转换成电能的发电系统,它的主要部件是由太阳能电池组、太阳能控制器、储能装置蓄电池(组)和太阳跟踪控制系统组成。其特点是高可靠性、寿命长以及对环境不产生污染、能独立进行发电且并网运行,受到世界各国电网公司的喜欢,发展前景十分广阔。太阳电池的发电功率通过“分流调节”全部变换为母线功率,一部分直接给负毂供电,另一部分则通过“充电调节”变换为充电功率为储能装置蓄电池组充电;蓄电池组功率通过“放电调节”变换为母线功率。对太阳电池发电功率的使用优先级依次为供电、充电、分流。充电功率可以视作母线的可调负载。太阳能电池光伏电源发电系统工作原理如图1所示。正丹线充电控制放电调节负载太阳能电池太阳能电池分流控制蓄电池组充电阼供电阵负母线图1光伏电源发电系统工作原理双向DC-DC转换器是连接正负母线电压与储能系统(如储能装置蓄电池组)的关键,所以使转换器的效率变髙极其重要。本文提出了一种降低功耗,提高整机效率的方案,使得对双问DCDC转换器的探讨变得更加具有意义。1.3国内外研究和应用现状20世纪后期,太阳能电池阵-储能装置蓄电池组构成的光伏电源发电系统的休积和重量庞大,著名外国学者提出了一种基于BCK/B0OST双向DCDC直流转换器来代替原有光伏电源发电系统的允电、放电模块,从而实现电压的稳定20世纪90年代,中国工程院院士陈清泉教授将基于BUCK/ BOOST双向DC-DC变换器在电动车领域使用,同年,外国专家研制了用大功率的水冷式DC-DC变换器即基于BUCK/ BOOST双向DC-DC直流转换器来驱动电动车,由于基于BUCK/BO0ST双向DC-DC变换器的输入输出电压的忙负极相反,不适合在电动车上应用,因此,他提出了一种基于BUCK-BO0ST级联型的双向DC-DC变换器,变换器的电源输入端与电压输出端的负端共用。经过4年时间,美国著名大学-弗吉尼亚大学教授李泽元开始研究在燃料电池上双向DC-DC变换器的配套应用。由此可见,用于载人航天的航天器电源和电动车辆的技术更新对双向DC-DC变换器的发展具有巨大的推动作用,随着开关直流变换器技术即脉宽调制技术的实现,给双向DCDC变换器的发展带来了曙光。1994年,有一位著名的澳大利亚学者发表论文,总结出几种非隔离型双向DC-DC变换器拓扑结构,主要是在CM0S开关管上反向并联具有快速、低功耗的二极管,且在二极管上反并联CMOS开关管,从而构成非隔离方式的双向DC-DC变换器种类有:BUCK-B0OST变换器、BUCK/B0OST变换器、CUK变换器和SEPI-ZETA变换器2004年,由我国学者张方华博士对推挽正激移相式、级联式、正反激组合式双向DC-DC直流变换器做了深入的研究。提出∫很多新型的应川电路,研究∫其控制模型,采用PI补偿环节的单电压闭环实现了系统闭环稳定。双向DC-DC变换器的硏究是近年来开关电源技术研宄的一个热门话题。2006年梁永春博士探讨了由反激式并联输入、串联输出构成的反激逆变器,提出了种同步整流的控制方案,极大地简化了髙频链逆变器的控制,使得整流二极管的导通损耗大幅度降低,整个电源系统的效率提高到85.8%。1.4论文主要的研究内容要求:设计一种双向DC-DC变换器,实现电池组的充电、放电功能。系统结构如图2所示,电池组由5节18650型、容量2000~3000mAh的锂离子电池串联组成。所用电阻阻值误差的绝对值不大于5%辅助电源测控电路3BS1 Rs-5Q2电双向DCDC池变换电路组RL=302直流稳压电源图2电池储能装置结构框图1.基本要求接通S、S3,断开S2,将装詈设定为充电模式(1)U2=30V条件下,实现对电池恒流充电。保障充电时电流l在1~2A范围内能够步进可调,步进值应≤0.1A,电流的控制精度≥5%。(2)设定1=2A,调整直流稳压屯源输出电压,使U2在2436V范围内变化时,要求充电电流I的变化率不大于1%(3)设定l1=2A,在U2=30V条件下,变换器的效率n1≥90%(4)测量并显示充电电流,在I-1~2A范围内测量精度不低于2(5)具有过充保护功能:设定l1=2A,当U1超过阈值U=24±0.5V时,停止充电。2.发挥部分(1)断开S1、接通S2,将装置设定为放电模式,保持U2=30±0.5V,此时变换器效率n2≥95%(2)接通S1、S2’断开S3’调整直流稳压电源输出电压,使直流电源电4压U在32~38V范围内变化时,双向DC-DC变换器能够自动切换工作模式即可自动切换充放电模式并保持输出电压U2=30±0.5V。(3)在满足要求的前提下简化结构、减轻重量,使双向DC-DC变换器、测控电烙与辅助电澒三部分的总重量不大于500g。(4)其他第二章双向Dc-D变换器拓扑结构的研究2.1双向DCDc变换器的基本原理与类型2.1.1双向DC-DG变换器的基本原理双向DC-DC变换器是把育流电压转换成另一个数值的电压,它是由软件控制导通的CW0S开关管、储能电感、续流二极管、具有滤波作用的电容、负毂等构成的,通过具有滤波功能的负载电路和直流电压时而使开关管时而接通或者时而关断,仗得另一端即负载上得到另一个直流电压2.1.2D0DG变换器的类型目前,国内外将双向DCDC变换器的拓扑结构主要划分为非隔离式和隔离式两大类。非隔离型拓扑的主要有:BUCK降压式、 BOOST升压式、BUCK- BOOST升降压型等拓扑。非隔离型拓扑如图3所示。隔离型拓扑的主要有:止激、反激、推挽、半桥、全桥型变换器(1)隔离型变换DYYYCD(a)BUCK变换器拓扑(b) BOOST变换器拓扑DL(c)BUCK- BOOST变换器拓扑图3非隔离型变换器拓扑以最基木的BUCK降压式变换器和BO0ST升压式变换器为例,介绍其工作原理。BUCK降压式变换器:当CMOS开关管Q接通时,电源Vin通过电感L给电容C充电;当开关管断开时,电感L通过快速、低功耗二极管D续流,电压逐渐降低。此时,电容上的电流由正逐渐降为零,最后变成负向,进而使开关管又一次导通,使得电感上电流增加。其储能电感L上电流波形如下图4所示tImar1-min(a)BUCK电感电流连续时波形(b)BUCK电感电流断续时波形图4BUCK电感电流波形BO0ST升压式变换器:当开关管Q导通吋,电源向电感L储能,电感L电流增加,负载由电容C供电;当开关管Q关断时,电感电流减小,电感电势与输入电压叠加,迫使二极管D导通,一起向负载供电,并同时向电容C充电。其电感电流波形如图5所小7
- 2020-12-05下载
- 积分：1
-
麻雀搜索算法（SSA）优化bp网络.rar
【实例简介】可以运行的代码！麻雀搜索算法（SSA）优化BP神经网络，能够很好的进行预测，并且该算法是今年提出的，非常好用，值得推荐和写论文
- 2021-11-30 00:33:20下载
- 积分：1
-
CLBP人脸识别程序及运行结果
matlab程序，包含了基本LBP，旋转不变LBP，均匀LBP ,旋转不变均匀LBP，CLBP等的LBP算子提取和分类，基于AR的人脸库，还有附加的一个说明文档以及自己测试的一些结果分析。
- 2020-12-05下载
- 积分：1
-
abaqus script reference manual
ABAQUS脚本参考文件，适合二次开发，ABAQUS各个模块的语句参考ABAQUS Offices and RepresentativesABAQUS, IncABAQUS Europe BVRising Sun millsGaetano Martinolaan 95166 Valley streetP.O. BOx 1637Providence RI 02909-24996201 BP MaastrichtTel:+14012764400The netherlandsFax:+14012764408Tel:+31433566906E-mail:support(@abaqus.comFax:+31433566908http://www.abaqus.comE-mail:info.europe@abaqus.comSales, Support, and ServicesUNITED STATESABAQUS Central, IncABAQUS East LLC1440 Innovation place300 Centerville road. Suite 209WWest Lafayette, IN 47906-1000Warwick. RI 02886-0201Iel:+17654971373Tel:+14017393637Fax:+17654974444Fax:+14017393302E-mail:support(@abaqusCentral.comE-mail: support@ abaqus EastcomAbAqUS Erie. incABAQUS Great Lakes, Inc3601 Green Road. Suite 31614500 Sheldon road. Suite 160Beachwood. OH 44122Plymouth, MI 48170-2408Tel:+12163781070Tel:+17344510217ax:+1216378107F7344510458E-mail:support(@abaqusErie.comE-mail:support(@abaqusGreatlakes.comABAQUS South, IncABAQUS West, Inc3700 Forums Drive Suite 10139221 Paseo Padre Parkway Suite FFlower Mound. TX 75028remont, CA 94538-1611Tel:+12145131600Tel:+15107945891Fax:+12145131700Fax:+15107941194F-mail:support(@abaqusSouth.comF-mail:supportabaqusWest.comARGENTINAAUSTRALIAKB Engineering S.R. LWorley Advanced analysisFlorida 274-Oficina 35Level 17. 300 Flinders Street1005 Buenos airesMelbourne Vic 3000ArgentinaTel:+61386125132Iel:+541143269176/7542Fax:+61392050573Fax:+541143262424E-mail:abaqus@worley.com.auE-mail: sanchezsarmientoarnet com arAUSTRIABENELUXABAQUS AuStria GmbHABAQUS Benelux EⅤZinckgasse 20-22/2/13Huizermaatweg 576A-1150 Vienna1276 LN HuizenaustriaThe NetherlandTel:+4319291625-0Tcl:+31355258424Fax:+4319291625-20Fax:+31355244257E-mail: support @abaqus.atE-mail: support(abaqus. nlCHINACZECH REPUBLICABAQUS ChinaSynerma s.r.oRoom a-2703 Eagle plazaHuntirov 58No. 26 Xiao yun rd468 22 SkuhrovBeijing, 100016Czech republicP.R. ChinaTel:+420603145769Tel:+860184580366Fax:+420603181944Fax:+860184580360E-mail: abaqus asynerma czE-mail:abaqus(@abaqus.com.cnFRANCEGERMANY(Aachen)ABAQUS France SasABAQUS Deutschland gmbH7 rue jean MMermozBat. aTheaterstrabe 30-3278000 VersaillesD-52062 AachenIel:+330139241540lel:+49241474010Fax:+330139241545Fax:+492414090963E-mail: support (@abaqus. frEmail: info(abaqus.deGERMANY(Munich)INDIA(Chennai)ABAOUS Deutschland gmbhABAQUS Engineering India(Pvt. LtdSendlinger-Tor -platz 83M. Prince arcadeD-80336 Munchen22-A Cathedral roadTel:+498959991768Chennai. 600086Fax:+498959991767Tel:+914428114624E-mail: info(abaqus.deFax:+914428115087E-mail: abaqus(@abaqus.coinITALYJAPAN(Tokyo)ABAQUS Italia s r IABAQUs. IncVia domodossola. 173rd Floor, Akasaka Nihon Building20145 Milano(MD)5-24, Akasaka 9-chome, Minato-kuTel:+390239211211Tokyo,107-0052Fax:+390231800064Tcl:+81354745817E-mail: info(@abaqus.itFax:+81354745818E-mail: tokyo(abaqus. jpJAPAN (Osaka)KOREAABAQUS, IncABAQUS Korea. Inc9th Floor, Higobashi Watanabe BuildingSuite 306, Sambo Building6-10. Edobori 1-chome Nishi-ku13-2 Yoido-Dong, Youngdeungpo-kuOsaka.550-0002Seoul.150-010Tel:+81648035020Tel:+8227856707Fax:+81648035021Fax:+8227856709E-mail: osaka @abaqus. jpE-mail: info(@abaqus. co kIMALAYSIANEW ZEALANDWorley advanced AnalysisMatrix Applied Computing Ltd19th Floor, Empire TowerP.O. Box 56-316. AucklandCity square centreCourier: Unit 2-5, 72 Dominion road Mt eden182 Jalan Tun razakauckland50400 Kuala LumpurTel:+6496231223Tel:+6032161226Fax:+6496231134Fax:+60321614266E-mail: abaqus-tech(amatrix co nzE-mail:abaqusmy@worley.com.auPOLANDRUSSIA, BELARUS UKRAINEBudSoft sp. z0.0TESIS Ltd61-807 PoznanOffice701-703,Sw Marcin 58/64l8. Unnatoy strIe:+486l8508466127083 Moscow. RussiaFax:+486l8508467Tel:+7095212-44-22E-mail:budsoft@budsoft.com.plFax:+7095212-42-62E-mail:infoatesis.com.ruSINGAPORESOUTH AFRICAWorley Advanced AnalysisFinite Element analysis Services (Ply)Ltd491B River valley roadUnit 4, The Waverley#09-01Ⅴ alley pointWyecroft roadSingapore, 248373Mowbray 7700Tel:+6567358444Tel:+27214487608Fax:+6567357444Fax:+27214487679E-mail:abaqus.sg@worley.com.auE-mail: feas(afeas co zaSPAINSWEDENPrincipia Ingenieros Consultores, SAABAQUS Scandinaviavelazquez, 94FEM-Tech ABE-28006 MadridPilgatan &cIel:+34912091482SE-72130 VasterasFax:+34915751026Tcl:+4621126410E-mail: abaqus(principia. esFax:+4621181244E-mail: abaqus abaqus.seTAIWANTHAILANDAPICWorley Advanced Analysis11F, No. 71, Sung Chiang road333 Lao Peng Nguan l BuildingTaipei. 1042820th Floor Unit bIel:+8860225083066Soi ChaypuangFax:+8860225077185Vibhavadi-Rangsit roadE-mail:apicaapic.com.twLadyao, JatujakBangkok 10900Tel:+6626893000Fax:+6626188109E-mail:abaqusth(@worley.com.auTURKEYUNITED KINGDOM(Cheshire)A-Ztech ltdABAQUS UK LtdPerdemsac Plaza, Teknoloji eviThe Genesis CentreBayar Cad, Gulbahar Sok., No: 17Science park south BirchwoodKozyataWarrington Cheshire WA3 7BH34742 IstanbulTel:+441925810166TURKIYEFax:+441925810178Tel:+902163618850E-mail: hotline abaqus. co ukFax:+90216361885E-mail:info(aa-ztech.com.tiSales OnlUNITED STATESABAOUS EaSt llc. mid-Atlantic officeABAOUS SOUth. Inc. Southeast office1 1 4 Zachary Court484 Broadstone wayForest hilL. MD 21050Acworth.GA30101lel:+14104208587lel:+17707950960Fax:+14104208908Fax:+17707957614E-mail:support(@abaquseast.comE-mail:support(@AbaqusSouth.comABAQUS West, Inc, Southern CA and Az Office ABAQUS West, Inc, Rocky Mountains Office1100 Irvine boulevard #2486910 Cordwood CtTustin Ca 92780Boulder co 80301Iel:+17147315895Tel:+13036645444Fax:+17142427002Fax:+1303200948lE-mail:inFo(aabaqusWest.comE-mail:Into(aabaqusWest.comFINLANDINDIA (Pune)ABAQUS Finland OyABAQUS Engineering Analysis Solutions(Pvt. LtdTekniikantie 12C-9. 3rd FloorFIN-02150 EspooBramha Estate, Kondwa roadTel:+358925172973Pune-411040Fax:+358925172200Tel:+912030913739F-mail: abaqus(@abaqus.seF-mail: abaqus( abaqus. co inUNITED KINGDOM(Kent)ABAOUS UK LtdGreat hollander business centre. Unit aMill Lane. UnderriverSevenoaks, Kent tN15 OSQTel:+441732834930Fax:+441732834720E-mail: hotline(abaqus. co ukPrefaceThis section lists various resources that are available for help with using ABAQUS, including technicalengineering and systems support, training seminars, and documentationSupportABAQUS, Inc, offers both technical engineering support and systems support for ABAQUS. Technicalengineering and systems support are provided through the nearest local support office. You can contactour offices by telephone, fax, electronic mail, the ABAQUs web-based support system, or regular mailInformation on how to contact each office is listed in the front of each ABAQUS manual. The ABAQUSOnline Support System(AOSS) is accessible through the MY ABAQUS section of the ABAQUS HomePage(www.abaqus.com).Whencontactingyourlocalsupportofficespleasespecifywhetheryouwouldlike technical engineering support (you have encountered problems performing an ABAQUS analysis orcreating a model in ABAQUS) or systems support (ABAQUS will not install correctly, licensing does notwork correctly, or other hardware-related issues have arisenThe ABAQUS Online Support System has a knowledge database of ABAQUS Answers. The ABAQUSAns wers are solutions to questions that we have had to answer or guidelines on how to use ABAQUS. Wewelcome any suggestions for improvements to the support program or documentation. We will ensure thatany enhancement requests you make are considered for future releases. If you wish to file a complaint aboutthe service or products provided by aBaqUs, refer to the aBaQus Home PageTechnical engineering supportABAQUS technical support engineers can assist in clarifying ABAQUS features and checking errors by givingboth general information on using ABAQUS and information on its application to specific analyses. If youhave concerns about an analysis, we suggest that you contact us at an early stage, since it is usually easier tosolve problems at the beginning of a project rather than trying to correct an analysis at the end.Please have the following information ready before calling the technical engineering support hotline, andinclude it in any written contacts:Your site identifier, which can be obtained by typing abaqus whereami at your system prompt(or byselecting Help-On Version from the main menu bar in ABAQUS/CAE or ABAQUS/Viewer)· The version of△ BAQUS that are you usingThe version numbers for ABaQUS/Standard and ABAQUS/Explicit are given at the top of the data(. dat)fileThe version numbers for ABAQUS/CAE and ABAQUS/Viewer can be found by selectingHelp-On Version from the main menu bar.The version numbers for the abaqus Interface for MoldFLOW and the abaQus Interface forMSC ADAMS are output to the screenThe version number for ABAQUS for CAtIa V5 can be found by selecting Help-AboutABAQUS for CaTiA V5 from the main menu bar in either of the aBaqus for CAtIA V5workbenchesThe type of computer on which you are running ABAQUSThe symptoms of any problems, including the exact error messages, if anyWorkarounds or tests that you have already triedWhen calling for support about a specific problem, any available abaqus output files may be helpful inanswering questions that the support engineer may ask youThe support engineer will try to diagnose your problem from the model description and a description ofthe difficulties you are having. The support engineer may need model sketches, which can be sent via fae-mail, or regular mail. Plots of the final results or the results near the point that the analysis terminated mayalso be needed to understand what may have caused the problemIf the support engineer cannot diagnose your problem from this information, you may be asked tosupply the input data. The data can be attached to a support incident in the aBaQUS Online SupportSystem. It may also be sent by means of e-mail, tape, disk, or ftp Please check the ABAQUS Home Page(http://www.abaquscom)forthemediaformatsthatarecurrentlyacceptedAll support incidents are tracked in the abaQUs Online Support System. This enables you( as well asthe support engineer) to monitor the progress of a particular problem and to check that we are resolving supportissues efficiently. To use the ABAQUS Online Support System, you need to register with the system. Visit theMY ABAQUS Section of the ABAQUS Home Page for instructions on how to register. If you are contactingus by means outside the aoss to discuss an existing support problem and you know the incident number,please mention it so that we can consult the database to see what the latest action has been and, thus, giveyou more efficient support as well as avoid duplication of effort. In addition, please give the receptionist thesupport engineers name if contacting us via telephone or include it at the top of any e-mail correspondenceSystems supportABAQUS SyStems support engineers can help you resolve issues related to the installation and running ofABAQUS, including licensing difficulties, that are not covered by technical engineering supportYou should install ABaQUS by carefully following the instructions in the ABAQUS Installation andLicensing Guide If you are able to complete the installation, please make sure that the product verificationprocedure was run successfully at the end of the installation procedure. Successful verification for licensedproducts would indicate that you can run these products on your computer; unsuccessful verification forlicensed products indicates problems with the installation or licensing (or both ) If you encounter problemswith the installation, licensing, or verification, first review the instructions in the abaQus Installation andLicensing Guide to ensure that they have been followed correctly. If this does not resolve the problems,consult the ABAQUS Answers database in the ABaQUS Online Support System for information about knowninstallation problems. If this does not address your situation, please create an incident in the AOSS anddescribe your problem, including the output from abaqus info=support. If you call, mail, e-mail, or faxus about a problem(instead of using the Aoss), please provide the output from abaqus info=support. Itis important that you provide as much information as possible about your problem: error messages from anaborted analysis, output from the abaqus info=support command, etc
- 2020-12-10下载
- 积分：1
-
仿网易新闻客户端源码+服务器
仿网易新闻客户端源码+服务器
- 2020-11-30下载
- 积分：1
-
易语言强制锁机源码
无法进入安全模式 强制锁系统 重启要输入你设置的开机密码 可按要求修改电脑用户名 请修改好源码后再使用 出现任何情况与作者无关！！
- 2021-05-06下载
- 积分：1
-
改进模拟退火算法求解TSP问题
改进模拟退火算法求解TSP问题也就是常说的旅行商问题
- 2020-11-30下载
- 积分：1
-
基于实例和特征的迁移学习算法研究_戴文渊
基于实例和特征的迁移学习算法研究上海交通大学学位论文原创性声明本人郑重声明:所呈交的学位论文,是本人在导师的指导下,独立进行研究工作所取得的成果。除文中已经注明引用的内容外,本论文不包含任何其他个人或集体已经发表或撰写过的作品成果。对本文的研究做出重要贡献的个人和集体,均已在文中以明确方式标明。本人完全意识到本声明的法律结果由本人承担。学位论文作者签名:日期:200年,月6日上海交通大学学位论文版权使用授权书本学位论文作者完全了解学校有关保留、使用学位论文的规定同意学校保留并向国家有关部门或机构送交论文的复印件和电子版,允许论文被查阅和借阅。本人授权上海交通大学可以将本学位论文的全部或部分内容编入有关数据库进行检索,可以采用影印、缩印或扫描等复制手段保存和汇编本学位论文保密□,在年解密后适用本授权书。本学位论文属于不保密囝。(请在以上方框内打“√”)学位论文作者签名:指导教师签名:1日期:209年,月日日期,亻年,月,b日上海交通大学学位论文答辨决议书姓名戴文渊学号1060394所在学科计算机应用技术指导教师俞勇答辩2009年1月答辩地点逸夫科技楼311日期16日论文题目基于实例和特征的迁移学习算法研究投票表决结果:555(同意票数/实到委员数应到委员数)答辫结论:回通过口未通过评语和决议:戴文谢的论女对基家例和将而习年译入的研鉴该课题旨右桌砖传纹戒器动限劫布衣器导成眼刺张采明意和有吃作包提了-种基于泉例汪移羽法给生了aAB算回提3-种基特公汪移3方,3因片和受本键技术取場3很山敌果,流文点明方法新颖,结沉正别,反强着己孔家地推第术基础理和相关的去步知议,具级的去从季科研午W彩机答排刷凶滴蕤,回间趣正确浴拜委员会讨(孔记名拨票奉决),-敌其通过硕士怪沉辩被子召丽士209年1月16日职务姓名职称单位签名主席黄林鹏教授上海交通大学答委员辩翁惠玉副教授上海交通大学委委员薛贵荣副教授上海交通大学员会委员张冬茉副教授上海交通大学成员委员俞勇教授上海交通大学签|天姿号委员秘书韩定助理研究员上海交通大学中文摘要基于实例和特征的迁移学习算法研究摘要传统的机器学习假设训练数据与测试数据服从相同的数据分布。然而,在许多情况下,这种同分布假设并不满足。不满足同分布假设的情况往往发生在训练数捃过期,而标注新数据非常昂贵。于是,我们有有了大量的在不同分布下的过期训练数据。完全丢弃这些数据将会是非常浪费的。在这种情况下,迁移学习就变得非常重要了,因为迁移学习可以从现有的数据中迁移知识,用米帮助将米的学习迁移学习( Transfer Learning)的目标是将从个环境中学到的知识用来帮助新环境中的学习仼务。因此,迁移学习不会像传统忛器学习那样作同分布假设。在本文中,我们将会比较全面的回顾迁移学习的整个领域,并且介绍我们在迁移学习领域的近期研究成果。我们的工作可以分为两部分;基于实例的迁移学习和基于特征的迁移学习。我们将会展示出,基于实例的迁移学习有更强的知识迁移能力,而基于特征的迁移学习具有更广泛的知识迁移能力。这两种方法各有千秋。我们介绍了两种迁移学习方法,分别基于 boosting技术和特征翻译。这两种算法分别对应基于实例的迁移学习和基于特征的迁移学习。我们通过非常全面的实验来证明我们的方法在迁移学习时候能够很大幅度提高很多现有的学习算法,无论是近迁移还是远迁移关键词:迁移学习、实例、特征英文摘要Instance-based and Feature-based Transfer LearningABSTRACTTraditional machine learning techniques make a basic assumption that the training andtest data should be under the same distributions. However, in many cases, this identicaldistribution assumption docs not hold. The violation of thc assumption might happen whenthe training data are out of date, but new data are expensive to label. This leaves plenty oflabeled examples that are under a similar but different distribution which is a waste throwaway entirely. In this situation, transfer learning becomes important to take the role of leveraging these existing data knowledgeTransfer learning aims at using learned knowledge from one context to benefit fur-ther learning tasks in other contexts. Thus, transfer learning does not make the identical-distribution assumption as tractional machine learning algorithms. In this thesis, we broadreview the whole field of transfer learning and then introduce our recent work on transferlearning accordingly. Our work can be divided into two parts: instance-based transfer learning, and feature-based transfer learning. We will show that instance-based transfer learninghas better strength in knowledge transferring, while feature-based transfer learning is withmore gerWe present two transfer learning algorithms based on boosting technique and featuretranslation respectively. These two algorithms corresponds to instance-based and featurebased transfer learning. Our extensive experiments show that our algorithms can greatlyimprove several state-of-the-art algorithms in the situation of transfer learning, includingnear transfer and far transferKEY WORDS: Transfer Learning, Instance, Feature目录目录摘要ABSTRACT(英文摘要)第一章绪论第二章迁移学习领域的研究现状2.1多任务学习.22跨领域学习23不同数据分布下的学习24其他迁移学习问题25迁移学习的应用4555678第三章基于实例的迁移学习算法研究3.1基于 Boosting的迁移学习算法3.1.1基本思想.3.1.2问题定义103.1.3 TrAdaboost算法描述..123.2 TrAdaboost算法的理论分析.14321基本符号143.2.2辅助训练数据上的错误率...:.···153.2.3源训练数据上的错误率,,203.3实验分析:··..··:·.:·.:···:·:·:·:······:213.3.1数据描述..22332实验结果23第四章基于特征的迁移学习算法研究254.1基于特征翻译的迁移学习算法4.1.1翻译学习框架274.1.1.1问题定义····‘·274.1.1.2风险最小化框架.41.1.3模型估计.··2941.14翻译器o304,2实验分析324.2.1实验数据酯鲁鲁D着着·,,·,,音唐鲁帝着争324.2.2比较方法324.2.3实验结果··········:····:..:.::.:::33第五章总结与展望35参考文献致谢4·,个人简历、在学期间的研究成果及发表的论文春,音42插图插图1-1日常生活中的迁移学习例子3-1关于 TrAdaboost算法基木思想的一个直观的示例。3-2 TrAdaboost算法的机制123-3一个关于数据生成的示例223-4三种算法在 people vs places数据集上的效果243-5 TrAdaBoost算法在 people vs places数据集上的达代曲线244-1一个直观的例子,用来说明六中学习策略的异同264-2共同出现数据的例子:Flickr(http://www.flickrcom/)74-3在12个数据集上的平均错误率3444对于不同的入, TLRLSK在12个数据集上的平均错误率34表格表格3.120 Newsgroups数据分布的描述2232SRAA数据分布的描述2333当只有1%的源数据是训练数据时的分类错误率4.1文本辅助图像分类的数据描述3
- 2021-05-06下载
- 积分：1
|
联系客服（QQ：3324461878）
