[Papermodels@emule] [Maly Modelarz 1973-02] - Polish Destroyer Burza

合集下载

SAE 1999-01-3279

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A.Tel: (724) 776-4841 Fax: (724) 776-5760SAE TECHNICAL PAPER SERIES SAE 1999-01-3279JSAE 9938034Analysis of Motorcycle Structural-Resonance-Induced Fatigue ProblemsLeRoy Petrick and Peter D. GunnessMTS Systems Corp.Reprinted From: Proceedings of the 1999 SAE Small Engine Technology Conference(P-348)Small Engine TechnologyConference and ExpositionMadison, WisconsinSeptember 28-30, 1999SAE 1999-01-3279 / JSAE 9938034 Analysis of Motorcycle Structural-Resonance-InducedFatigue ProblemsLeRoy Petrick and Peter D. GunnessMTS Systems Corp. Copyright © 1999 Society of Automotive Engineers, Inc.ABSTRACTVehicle structural resonance modes are classified gener-ally into rigid and flexible (non-rigid) body modes. During motorcycle testing and development for design validation, it is often useful to understand these modes of vibration. Understanding rigid and flexible body modes helps to improve the ride and handling performance. Understand-ing the flexible body modes helps to isolate noise, vibra-tion, and harshness (NVH) problems. It can also help to find the root causes of structural durability failures. Flexi-ble body modes can also be annoying or unsafe to the operator. For example, handlebar vibrations may cause numbness in the hands or arms. Flexible body modes also can contribute to motorcycle dynamic instability modes such as the weave instability. Similarly, the rider’s ability to see approaching traffic from the rear may be reduced if mirrors are vibrating due to a flexible body mode in the handlebars, frame, or front fork. Therefore, experimental measurement and analysis of the structural resonance modes of motorcycles helps to assess prob-lematic mechanical vibrations. Continuous exposure to mechanical vibrations can cause both human fatigue and structural fatigue. This paper focuses on the assessment of structural-resonance-induced fatigue problems. INTRODUCTIONThe animation of structural modes using operational deflection shapes (ODS) can help to quickly gain an understanding of significant motorcycle structural reso-nance modes. Engine order tracking when used in com-bination with ODS can be used to distinguish if the flexible body modes are road induced or engine induced. Vehicle ride and handling can be analyzed using dynamic analysis. Spindle forces and moments can also be con-verted to body forces and moments using modeling tech-niques. Finite element analysis can be used to convert loads into stress or strain. Fatigue analysis of strain data can be used to quantify whether engineering design changes have lasting beneficial durability results.THEORY—MODAL ANALYSIS VS. OPERATIONAL DEFLECTION SHAPE ANALYSIS A structure can be mathematically modeled by dividing it into discrete masses connected by discrete stiffnesses. The masses and stiffnesses are known, respectively, as the structural mass and the structural stiffness of the modeled structure. An equation that balances the forces for each of the masses can be written using Newton’s Second Law. These equations can be organized in matrix form to aid in their solution. This matrix form of the equa-tions is known as the equations of motion. The mass and stiffness distribution of the structure is on one side of the equations and the forces applied to the structure are on the other side of the equations.If the forces applied to the structure are set equal to zero, the solution of the resulting equations is referred to as modal analysis. The solution of this form of the equations of motion yields a set of frequencies and associated deformation shapes. In a complete solution of the Equa-tions of Motion there are as many frequencies and shapes as there are masses in the model. However, most practical solution methods do not compute (extract) all the possible frequencies and shapes. These frequencies are known as Eigen values or natural frequencies or mode frequencies. The shapes are known as Eigen vec-tors or mode shapes. These frequencies and shapes are the modal properties of the modeled structure. The modal properties are an alternative form of the structural mass and stiffness.Experimental modal analysis is an empirical method of solving the equations of motion. In this method, an inte-gral transform known as the Laplace transform is be used to solve the equations of motion. The Laplace transform of the equations of motion results in a matrix of functions known as the transfer function (TF). The procedure to empirically solve the equations of motion is to derive the TF from a measured frequency-response-function (FRF). The Fourier transform is a special case of the Laplace transform and the FRF is a special case of the TF. This solution procedure uses a process know as parameter estimation or curve fitting to extract the modal properties from the measured FRF.If the forces applied to the structure are not set to zero, a form of the experimental modal analysis procedure can still be applied. This form of the procedure is referred to as operational deflection shape (ODS) analysis. The results are not modal properties and are therefore not an alternative form of structural mass and stiffness. The ODS results represent one frequency component of the total response. The response of the structure is the result of the applied forces. The applied forces can be known or unknown. Since ODS represents a single component of a complex motion and contains the product of the Struc-tural Mass/Stiffness and the Applied Forces, it is valuable in understanding the relationships between mass/stiff-ness distribution and applied forces.ODS BACKGROUND AND PURPOSEODS is a tool that can be used in conjunction with labora-tory road simulation. The servohydraulic simulators used for full motorcycle testing may be used to excite the spec-imen using either single-channel excitation or multi-chan-nel random orthogonal excitation. Servohydraulic actuators have been proven to be very repeatable, and when combined with frequency-shaped and scaled ran-dom data, which is designed to be periodic, several prob-lems of studying road induced mechanical vibrations are easily overcome. The inherent repeatability of the servo hydraulic system allows one to play out a multi-channel drive file many times while each time collecting a different set of acceleration location measurements. The orthogo-nal drive file makes it possible to understand the effects of different excitation sources. The same software used to calculate a system FRF to simulate proving ground roads in the laboratory can also be used to prepare data for ODS analysis. A major difference between modal analysis and ODS is that the modal amplitude term is the ratio of acceleration per unit of force, while the ODS amplitude is the ratio of an acceleration signal to another acceleration signal. Data may be either collected on the road and post processed in the laboratory, or collected in the laboratory using random road excitation. There are inherent advantages of ODS measurements in the lab:•Data is easier to collect in a laboratory setting.•System repeatability allows multiple measurements to be treated as a single set.•The excitation source is designed to be periodic in an FFT (fast Fourier transform) frame.•Orthogonal data makes it easy to understand causal-ity in multi-axial systems.•The spectral shape and excitation level can be easily modified to reflect various road surfaces.In this study, two vertical actuators were used for ODS simultaneous excitation. Two spindle acceleration signals were used as the ODS reference transducers for 22 triax-ial acceleration measurements. The ODS technique could also be applied to repeatable mechanical test sys-tems with more than two channels of excitation.ODS CASE STUDYIn this study, a prototype motorcycle tested on a tire cou-pled road simulator developed a premature failure in the chassis. Proving ground failures had not occurred on similar specimens, but the simulator is capable of dupli-cating fatigue damage at a rate of several hundred miles per hour when compared to on-road testing. Therefore, an investigation of the fatigue failure was conducted in parallel with the following:•Motorcycle laboratory simulation durability testing •Proving ground vehicle testing•Public road vehicle testing•Engineering design modifications•Prototype manufacturing process changesAn instrumented specimen, including strain gages in the area of concern, was placed on the two-post simulator. Data was collected from a variety of simulated public roads in the test lab. Data was analyzed in the time domain and the frequency domain, and the fatigue dam-age was also calculated. Frequency analysis indicated a trend for the strain gages to be subject to a resonance. (See Figure 1.) Several specimen structural resonance modes could be observed while the specimen was on the simulator. However, it was not apparent whether the front or rear actuator had a larger influence on these modes. To help clearly understand the structural resonance modes affecting this region of interest, ODS analysis was applied, to concisely summarize the vehicle’s operational deflection shapes.To complete the ODS exercise, a prototype motorcycle was placed on a four-channel spindle-coupled simulator. Although this simulator was capable of using two vertical and two longitudinal inputs, only the two vertical actua-tors were used for ODS. The front longitudinal strut was disconnected to allow wheelbase changes due to vertical excitation. This setup most closely resembles the con-straints on the road.The ODS measurements were made by playing out the same multi-channel orthogonal random drive file nine times (see Figure 3). Prior to each of the multiple mea-surements, an array of eight linear accelerometers were moved to the pre-defined 22 locations. This random drive file is sent as an external command to a servo-hydrauli-cally controlled mechanical test fixture. Three repeats at each reference acceleration location were used to con-firm that the excitation conditions remained consistent (see Figure 4). The spatial location of the 22 frame loca-tions and two reference locations were input as X, Y, and Z coordinates in the software used for ODS analysis. The ODS analysis indicated a mode of structural reso-nance at the same frequency as the peak in the auto spectrum of the strain data. (See Figure 5.) This mode corresponds to the fork bending mode of vibration. Both ODS and experimental observations indicate that both front and rear vertical excitation can induce front forkbending. While it is counterintuitive that the fork bending mode of vibration is excited more by the rear actuator than by the front actuator, it is possible that the rear input may cause more vehicle pitch and heave –hence fork strains may be resultant of forward pitch, upward heave and bump combined. (See Figure 6.) This resonance peaks at 16.8 Hz.Strain data, acquired near the area of interest, collected from simulated rough road events, was digitally filtered and analyzed using fatigue analysis to understand if there is a relationship between the vibration mode and the highly strained area. The original data and the data which has been band-pass filtered from 14–19 Hz show that a large amount of the strain gage signal content comes from the fork bending resonance. (See Figure 7.)Fatigue analysis was subsequently done on both data sets to help understand if structural vibrations are the root cause of highly strained areas on the chassis. (See Table 1.)The fatigue content of the measured simulation data con-taining 0.5–34 Hz frequency content is normalized to 100% in the table. Although roughly one-half of the total signal peak amplitude is contained in the 14–19 Hz range, it can be seen that the 14–19 Hz data has only 13% of the calculated damage of the entire signal. How-ever, 0.5–14 Hz and 19–34 Hz is also only 11% of the calculated damage of the entire signal. This is because real road signals are composed of various sinusoidal components which, having varying amplitude, frequency,and phase, add together to become more damaging than the sum of the individual components.CONCLUSIONODS is a useful tool for understanding vehicle structural resonance modes. When used in combination with other engineering trouble-shooting tools, it helps to visualize structural-resonance-induced fatigue problems. Fatigue analysis can be used in conjunction with ODS to improve the structural durability of prototype motorcycles.CONTACTMTS Systems Corporation Test Consultants or Motorcy-cle Market Manager for further information:leroy.petrick@ peter.gunness@ gary.stewart@ Figure 1.ASD of Steer Head Strain DataFigure 2.Typical Four-channel Road Simulator Used for ODS Testing. (Motorcycle shown is different from that used in study.)Figure 3.Multichannel Orthogonal Random SignalTable 1.Frequency Damage % Damage0.5-340.016128100.00%14-190.00213913.26%0.5-14, 19-350.00175310.87%Figure 4.Repeatability Error vs Signal LevelFigure 5.ODS Triaxial Acceleration MeasurementPointsFigure 6.16.8 Hz ODS AnimationFigure 7.Unfiltered vs. Band-Pass Filtered Road Data。

NUVOTON NuCam 硬件用户手册说明书

NNuvot The info Nuvoton Tech ton is providinFor addition Nu Hard formation desc hnology Corpo ng this docum design. Nuvo All data andnal informatio N329uMak dwar cribed in this oration and sh ment only for r oton assumes d specification on or questionw AR 905 F ker Nu re Use document is hall not be repreference purp s no responsib ns are subjecns, please con www.nuvoton.cRM ® AR 32-b amily uWica er Ma the exclusive produced with poses of NuM bility for error ct to change w ntact: Nuvoto comRM926E it Micro yam anuale intellectual p hout permissio Micro microcon rs or omission without noticen Technology EJ-S B oproce property ofon from Nuvo ntroller based ns..y Corporationasedessoroton.d system.Table11.122.12.22.32.42.52.62.62.62.62.72.82.82.82.82.92.1033.13.23.33.43.5456e of ConOVERVIEWNuWicaIntroductioNuWicaN32905GC030TVP515RTL818Board I6.1 NuWi6.2 NuWi6.3 NuWiNuWicaNuWica8.1 NuWi8.2 NuWi8.3 NuWiN32905NuWicaNuWiCamNuWicaNuWicaNuWicaNuWicaNuWicaStarting toExample CREVISIONntentsW..........am Systemon to NuWiam Key Fea5R3DN on N8 on NuWic50/GM715089FTV on NIntroductioncam-VGA Bocam-TVP Bocam-Debugam Power Sam Jumpercam-VGA Bocam-TVP Bocam-Debug5R3DN Pinam PCB PlaSchematicam-VGA Scam-VGA Scam-TVP Scam-TVP Scam-Debug SUse NuWCode.......N HISTORY............Block ......cam.......atures ......NuWicam Fcam Feature0 on NuWicaNuWicam Fen .............oard ..........oard ..........Board........Scheme ....and Conneoard Jumperoard JumperBoard JumpeAssignmenacement ...cs .........chematic (1/chematic (2/hematic (1/2hematic (2/2Schematic (WiCam .................Y.................................................................Features ....es ...........am Featureeatures .................................................................................ector .........r Description.Description.er Descriptiont for Conne.........................../2) .........../2) ...........2) ...........2) ...........(1/1)................................................................................................................................s....................................................................................................................................................on .............ectivity .................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................. (3) (3) (4) (4) (5) (6) (7) (8) (9).. 10.. 11.. 12.. 12.. 13.. 13.. 15.. 17.. 18.. 21. 24.. 24.. 25.. 26.. 27.. 28. 29. 30. 311 O Nu vidTh de funTh rem theTh TX de exN 1.1Figex OVERVIEW uvoton’s NuM deo and audi he NuWicam ecoder, voice nctions make he NuWicam motely. The e Wi-Fi came he NuWicam XD/RXD prot evice and Ar xpand various NuWicam S gure 1-1 shoxternal device WMaker NuWic io data to a w integrates n e microphon e a cost effec can be use NuWicam p era easily and m has an ext ocol of UART rduino microc s possible ap System Bl ows the NuWes. Key part f cam is an op wireless rece necessary fu ne, SDIO ho ctive solution ed for home rovides an A d connect to ternal conne T for commu controller or pplications. ockWicam syste features in NFigure 1-1pen source hiver through unctions such ost controlle for Wi-Fi ca security or m App supports it from every ector that pro unication, suc for Nuvoton em architect NuWicam are NuWicam Sardware devWi-Fi.h as CMOS er for conne mera applica monitoring th s iOS and An ywhere throu ovides Ardui ch as do som n’s NuEdu-SD ture and per also listed.System Block vice supports sensor or T ecting with W ations. he elders, ba ndroid that a gh a smartph no microcon me data sam DK- M451 d ripheral interks P2P that traTV video (NT Wi-Fi moduleabies or pets allows users hone or table ntroller based mpling betwee development rfaces for coansmits aTSC/PAL) e. Theseat hometo set up et.d kit withen mobileboard to onnecting2 INThN3SDCMdifsptwmiN2.1●●●●●●●●●●●●●●NTRODUChe NuWicam32905R3DNDIO Wi-Fi mMOS sensorfferential micpeed device pwo UART poicrocontrollerNuWicamSOC: N3Main proVGA CMAlternativVideo strSDIO WiEmbeddeSupports3 LEDs foWide powSystem oMain PCBDebug to⏹ Softport⏹ MesPower co⏹ A/V⏹ WorTION TO Nm uses anprovides themodule for wor TV videocrophone intport to PC foorts, port 0r based kit coFeatures32905R3DN (gram with 8OS Image Sve video withreaming achi-Fi module wed Electret Cs Arduino micor status indiwer supply inoperation in LB size: 25mmooling board,tware develotssage debugonsumptionstreaming rurking time ofNUWICAMARM926 Se SPI interfaireless netwoo (NTSC/PAterface withr software deis dedicateommunication(ARM926 atMB SPI Flasensor (GC03h TV video deeves MJPGwith RTL8189Condenser Mcrocontrollerication, Prognput range froLinux OS 3.4m x 50mmwhich is useopment or SPging with UAunning up to6 hours is poMOC whichce for SPI Nork connectiL) decoder fECM type fevelopment oed for debun.192 MHz, 32sh (W25Q64F308)ecoder, TVP5performance9FTV,softwaricrophonecommunicatiram Ready, Wom 3.5V to 6V4.35ed for:PI flash progrART port 0 th200mA at 5Vossible with 1N32905R3DNOR flash boion, providesfor Motion JPfor audio stror SPI flash’sugging and2 MB DDR2 wFV, SO8)5150/ GM715e at VGA@30re AP mode sion based onWi-Fi Link &Vram updatedrough USB vV (typical)1500mAH AADN as the Sooting, provids a video inPEG video sreaming, pros program upthe UARTwith TQFP6450 (NTSC/PA0fpssupport andn TXD/RXD pApp runningby the utilityvirtual COMA BAT x 4SOC microcdes a SD intput interfacestreaming, provides an USpgrade. It alsport 1 for4 package)AL support)easy to confprotocolgof PC througcontroller,terface toe to VGArovides aSB2.0 Hio provideArduinoiguregh USBN 2.2●● ● ● ● ●● ● ● ● ●PlN32905R3D N32905R CPU ope SRAM@System bCCIR601Resolutio video stre DMA acc 10-bit Au USB2.0 H Two UAR Package ease refer to DN on NuW R3DN: Nuvot erated at 1928K and Inter booting with S 1 & CCIR656on up to 2M p eaming celerate that udio ADC with Hi speed, driv RT ports supp , TQFP-64 (Mo Figure 2-1 f Wicam Fea ton’s ARM9 f MHz,1.8V w rnal Booting R SPI NOR flas 6 interfaces fo pixel for Still SD data tran h Microphone ver support M port debuggi MCP, stacke for the N3290Figure 2-1 N aturesfamily based with 8KB I-Ca ROM, IBR@sh or CMOS ima Image Captu nsfer with SD e pre-Amp &MS (Mass St ng and Ardu d with 32MB 05R3DN SON32905R3DN on the ARM ache & 8KB D 16Kage sensor ure, 640x480IO Wi-Fi mod & AGC torage) Class ino controller **********C outline.N SOC Outlin 926EJ-S cor D-Cache 0 (VGA) resol dule s r communica V)nere lution for MJPationPEGG2.3●●●●●●●PlGC0308 onGC0308:ResolutioOn-ChipOperatingOutput foSingle poSupport Hease refer ton NuWicam: GalaxyCoreon 640V x 48ISP providesg at up to 30ormats: YCbCower supply (Horizontal /Vo Figure 2-2 fFigurem Featurese’s CMOS im80H with 1/6.5s AE (Auto Eframes per sCr4:2:2(3V)Vertical mirrorfor the GC03e 2-2 GC030smage sensor5-inch opticaExposure) andsecond at 24r control308 CMOS se08 CMOS Seal formatd AWB (Auto4 MHz clock iensor modulensor Moduleo White Balanin VGA modee outline.e Outlinence) controleT 2.4●● ● ● ● ● ● ●● ●PlTVP5150/G TVP5150GM7150 Both vide Support N The outp Single 14Support p NTSC/PA ⏹ TVP ⏹ GM TVP5150GM7150 ease refer toGM7150 on 0: Ti’s NTSC/is Chengdu eo decoders NTSC (N, 4.4put formats ca 4.318-MHz C power-on res AL detection P5150 can su M7150, detect 0 PKG is LQF PKG is QFN o Figure 2-3 f Figure 2n NuWicam/PAL video d GoldTel (成都are pin out c 43), PAL (B, an be 8-bit 4:Crystal for all set upport auto d tion by jumpe FP32 N32for the TVP52-3 TVP5150/m Featuresdecoder 都国腾) NTSC compatible D, G, H, I, M :2:2 or 8-bit Istandards detectioner setting150/ GM715/ GM7150 TV C/PAL video M, N) Video d TU −R BT.6550 TV video dV video deco decoderata source56 decoder outlin oder outlinene.R2.5●●●●●●●PlRTL8189FTRTL8189OperateModulesOperateSecurity1T(transand 150NuWicamease refer toTV on NuW9FTV: Realted at 802.11nsize is 12.0 mat ISM 2.4Gcan apply Wsmitter) andMbps upstrem Network Ao Figure 2-94FigureWicam Feaek’s Wi-Fi chn Wi-Fi LGA Mmm x12.0 mmGHz frequencWPA/WPA2 c1R (receiveream PHY ratArchitecture w4 for the RTL82-4 RTL818turesipModulem x 1.6mm, 3cy bandscertification fo) allow data resworking at so8189FTV SD89FTV SDIO3.3V at SDIOor Wi-Fi.rates supportftware AP mDIO Wi-Fi moWi-Fi ModuleO Interfaceting up to 15odeodule outlinee Outline0 Mbps down nstreamB 2.6As shodebug●● ●Board Intro own in Figure board, as de NuWicam NuWicamNuWicam oductione 2-5, the NuW escribed belo m-VGA: Wi-F m-TVP: Wi-F m-Debug:De Wicam soluti ow: Fi camera ma Fi camera ma ebug boardFigure 2-5on contains t ain board witain board with5 NuWicam P two kinds of th VGA CMOh TV video dPCB Outlinemain boardsOS sensorecoders and one commmon2.6.1The NubasedstreamresolutNuWicam-uWicam-VGAon Nuvoton’sming over Wi-tion.-VGA BoardA board usess N32905R3Fi network wFigure 2-6ds GC0308 VGDN for encodwith RTSP seNuWicam-VGGA CMOS seding. NuWicarver. The forGA Board (Fensor and proam firmwaremat of videoFront View anovides a powprovides botstream is Mond Rear Viewwerful JPEG cth audio andotion-JPEG ww)codecvideowith VGA2.6.2 The Nu analog encodi server.NuWicam-uWicam-TVP camera and ng. NuWicam The format -TVP Board P board uses d based on N m firmware pof video streFigure 2-7dTVP5150 or uvoton’s N32rovides botham is Motion NuWicam-TV r GM7150 TV 2905R3DN toaudio and vn-JPEG withVP Board (FV video decod o provide a pideo streami VGA resolut ront View an der for conne powerful JPE ng over Wi-F ion. nd Rear View ecting NTSC EG codec for Fi network wit w)C or PALvideo th RTSP2.6.3The Nusoftwarfrom thN2.7Figurerequireboard dNuWicam-uWicam-Debre developmehe N32905R3NuWicam2-9 shows thement to provdamaged or-Debug Boabug board is cent or SPI fla3DN UART pFigure 2-8 NPower Schhe NuWicamvide that corrbroken.ardconnected toash’s programport 0 throughNuWicam-Dehemesystem powrect power voFigure 2-9o the NuWicam upgrade. Uh USB Virtuaebug Board (Fwer scheme, uoltage to NuWNuWicam Poam-VGA or NUser can alsoal COM.Front View auser should fWicam to preower SchemeuWicam-TVPo get the debnd Rear Viewfollow the poevent incorreceP main boardbugging messw)ower distributct voltage cad forsageionusedN2.82.8.1FigureNuWicam JNuWicam-2-10 showsFigure 2-10 N●CONmicr⏹Note:Fi camUSBprefe⏹Jumper an-VGA Boardthe front andNuWicam-VGN1: This connrocontroller bCON1. pin◆ PIN1◆ PIN2: The pin 1-2mera functionpower, abour and recommCON1. Pincommunicand Connecd Jumper Dd rear view ofGA Board & Nnector is an ebased kit1-2, it is for p, input voltag2, connect it t2 of CON1 isn with an indt voltage reqmended.3 & 4, UARTationVSS3.5V~6V,VIN Q2SDGctorDescriptionf NuWicam-VNuWicam-Deexpanded copower supplyge range fromto GNDconnected toependent poquest is fromT protocol ofHTXDHRXD, BAT/DC-SI2301CDSVGA Board &ebug Board (onnector for cy or BAT conm +3.5V to +6o power suppower supply w3.5V (Min.) tinterface wit& NuWicam-D(Front View aconnecting tonnection6Vply or BAT, Nwithout NuWito 6V (max.),th TXD/RXDDebug Boardand Rear Vieo ArduinoNuWicam canicam debug btypical DC 5for Arduinod.ew)n run Wi-board’s5V is● CON● SW1⏹N2: This conn1: RESET KESystem re-nector is forEY-start when thconnecting whe button is pwith NuWicampressedm-Debug boaard’s CON32.8.2Figure NuWicam-2-11 showsFigure 2-11●J1: ●J2: ⏹ ⏹Note jump ●CON kit⏹ Note:Fi cam USB prefe ⏹-TVP Board the front view NuWicam-TV NTSC/ PAL NTSC/ PAL v Jumper J2 Jumper J2 e: TVP5150 per J2 setting N1: This is an CON1. pin ◆ PIN 1◆ PIN 2: The pin 1-2mera function power, abou r and recomm CON1. Pin communicad Jumper De w and rear vi VP Board & N analog came video source open expresclose exprescan automat g. n expanded 1-2, is for po: input voltag 2: connect it t 2 of CON1 is n with an ind t voltage req mended. 3 & 4, UART ation escriptioniew of NuWic NuWicam-Deera input con e selection (th sses video so sses Video s tically identify connector fo ower supply o ge range from to GNDconnected to ependent po quest is from T protocol of cam-TVP Boa ebug Board (nector his jumper fu ource is NTS ource is PAL y NTSC/PAL r connecting or BAT conne m +3.5V to +o power supp ower supply w 3.5V (Min.) t interface wit ard & NuWic (Front View a unction only f C L L signals by fi to Arduino m ection 6V ply or BAT, N without NuWi to 6V (max.),th TXD/RXD cam-Debug B and Rear Vie for GM7150 u irmware itsel microcontrolle NuWicam can icam debug b typical DC 5for ArduinoBoard.ew)use)f without er based n run Wi-board’s 5V is●CON●SW1⏹N2: This con1: RESET KESystem re-nector is forEY-starts when tconnecting wthe button iswith NuWicampressed.m-Debug boa ard’s CON32.8.3 FigureNuWicam-2-12 shows ● J1: U ⏹ ⏹Note: For Guide.pdf● CON ⏹ ● CON ⏹ ● CON⏹N ote: ForNuWicam -Debug Boa the front viewFigure 2-1USB recover J1 jumper o J1 jumper c Flash progr r AutoWriter f”.N2: USB con Used for Umust conneN3:Connected.N4: VCOM co User can gUSB Virtuar Virtual COMm User Guide ard Jumper w and rear vi 12 NuWicam ry mode sele open, for nor close, N3290ram upgrade program ope nnectorSB applicatio ect USB cabl to NuWicam onnector et that debug al COMM driver insta e.pdf”.r Descriptio iew of NuWic -Debug Boarctionrmal operatio 05R3DN ente e in AutoWrite eration, pleas on or SPI Fla le through CO m-VGA or Nu gging messa allation and o ncam-Debug B rd (Front and on er USB recov er program.se refer to the ash program ON2 to PC fo Wicam-TV m age from N32operation, ple Board.d Rear View)very mode, th e “Nuvoton N upgrades, fo or getting US main board’s 2905R3DN U ease refer to his case only NuWicam Use or this purpos SB communic CON2ART port 0 t the “Nuvotony for SPIer se, usercation. hroughnN2.9The NufollowinPKG No123456789101112131415161718192021222324N32905R3DuWicam usesng table..# Pin NameGPD[13]GPD14]GPD[15]GPE[4]GPE[5]GPE[6]GPE[7]GPE[2]GPE[3]XINXOUTMVREFU_PLL_VUD_DMUD_DPUD_VDD3UD_REXTVSSGPC[15]GPC[14]GPC[13]GPC[12]GPC[11]GPC[10]DN Pin Ass N32905R3De FSSSSSSSSSXXMDD18 UUU33 UT UGSSSSSSsignment fDN (TQFP64FunctionSPI0_CS0_SPI0_DISPI0_DOSDDAT[2]SDDAT[3]SDCMDSDCLKSDDAT[0]SDDAT[1]XINXOUTMVREFU_PLL_VDD18UD_DMUD_DPUD_VDD33UD_REXTGNDSPDATA[7]SPDATA[6]SPDATA[5]SPDATA[4]SPDATA[3]SPDATA[2]for Connec4 with EPAD)ctivity) as SOC. AlConnectivitySPI FlashSPI FlashSPI FlashSDIO WI-FISDIO WI-FISDIO WI-FISDIO WI-FISDIO WI-FISDIO WI-FI12 MHz12 MHz0.9V1.8VUSBUSB3.3V12.1KGNDVGA / TVP5150VGA / TVP5150VGA / TVP5150VGA / TVP5150VGA / TVP5150VGA / TVP5150l pin definitioons are listed in theTypeIOUIOUIOUIOUIOUIOUIOUIOUIOUAAPPAAPAGIOUIOUIOUIOUIOUIOU25262728293031323334353637383940414243444546474849505152GPC[9]GPC[8]VDD18VDD33ADC_VSSADC_VDDADC_AINADC_AINADC_AINGPA[5]GPA[4]GPA[3]GPA[1]RST_GPD[1]GPD[2]UD_CDETVDD18MVREFVDDQADAC_HPADAC_HPADAC_HPADAC_HPADAC_AVADAC_VRADAC_AVVDD18SS13S33 AD33 AN[0] MN[1] MN[2] NTSSTRHHT U1MVPVSS33 APVDD33 APOUT_L NPOUT_R NVSS33 AREF NVDD33 A1SPDATA[1]SPDATA[0]1.8V3.3VADC_VSS33ADC_VDD33MIC+MIC-NCTBDSW_SCKSW_SDATBDRST_HUR_TXDHUR_RXDUD_CDET1.8VMVREFVDDQADAC_HPVSS3ADAC_HPVDD3NCNCADAC_AVSS33NCADAC_AVDD331.8V333VGA / TVP5150VGA / TVP51501.8V3.3V3.3VMICMICNCreservedI²CI²CreservedRESETto Arduino RXto Arduino TXUSB1.8V0.9V1.8VGND3.3VGND3.3V1.8VIOUIOUPPGPAAAIOUIOUIOUIOUIUIOUIOUIPPPGPAAGAPP53545556575859606162636465GPA[7]VDD33GPA[10]GPA[11]GPB[6]GPB[5]GPB[4]GPB[3]GPB[2]GPB[1]GPB[0]GPD[12]GNDS3UULLLSSSSSGS_RESET/RECO3.3VURTXDURRXDLED3_TBDLED2_LINKLED1_REDAYSVSYNCSHSYNCSPCLKSCLKO]/ NTSCSPI0_CLKGNDOVERYor PAL SELvideo device RE3.3VVCOM debugVCOM debugstatus LED, lowstatus LED, lowstatus LED, lowVGA / TVP5150VGA / TVP5150VGA / TVP5150VGA SCLK / TVSPI FLASHGNDESET/ USB recow activew activew activeVP5150 CAM soovery modeource selectionIOUPIOUIOUIOUIOUIOUIOUIOUIOUIOUIOUE_PA AD2.10FigureNuWicam2.14, Figure PCB Place 2.15 and FigFigure 2-1ementgure 2.16 sho3 Front andow the front aBack NuWicaand back Nuam-VGA PCuWicam PCBB Placement B placement.tFigure 2-14 Front andBack NuWiccam-TVP PCB BPlacementtFigure 2-155Front and BBack NuWica am-Debug PC C B Placemen nt3 NN3.1UWICAMNuWicam-SCHEMATVGA ScheTICSematic (1/2)N 3.2NuWicam-VGA Sche ematic (2/2)N3.3NuWicam-TVP Schem matic (1/2))N 3.4NuWicam-TVP Schem matic (2/2))N3.5NuWicam-Debug Sch hematic (1/1)4 STo useFirmwaTARTINGe NuWicam, pare programmTO USE Nplease refer tming, MobileNUWICAMto the “NuWiApp installatcam User Gution and Virtuuide.pdf” to gual COM drivget Hardwarever installatioe connectivityon informationy,n.5 EXAs to NserverXAMPLE CNuWicam exaconfigurationCODEample progran and A/V Stram, please rereaming oveefer to the “Nr RTSP inforuWicam progrmation examgramming gumple code.uide.pdf” to gget HTTP6 RD2REVISION HDate 2016.07.27 HISTORYRe 1.0evision00Descri 1. I iption Initially issued.NuvotomalfundamagInsecuenergydynamtypes oAll Insclaimsdamagon Productsnction or faige. Such appure usage iny control inmic, brake oof safety desecure Usags to Nuvotoges and liabs are neitheilure of whicplications arncludes, butnstruments,or safety sysvices, and oge shall ben as a resuilities thus iImr intended nch may causre deemed, “t is not limitairplane orstems desigother applicamade at cuult of custoncurred byportant Nonor warrantese loss of h“Insecure Uted to: equir spaceshipgned for veations intenustomer’s riomer’s InsecNuvoton.oticeed for usagehuman life, bsage”.pment for sp instrumentehicular useded to suppisk, and incure Usage,e in systemsbodily injurysurgical impts, the conte, traffic sigport or sustathe event th, customers or equipmy or severeplementationtrol or opergnal instrumain life.hat third pashall indemment, anypropertyn, atomicration ofments, allarties laymnify the。

Multi-modal

Multi-modal retrieval of trademark images using globalsimilarityS.Ravela R.ManmathaMultimedia Indexing and Retrieval GroupCenter for Intelligent Information RetrievalUniversity of Massachusetts,Amherst,MA01003Email:ravela,manmatha@AbstractIn this paper a system for multi-modal retrieval of trademark images is presented.Images are characterized and retrieved using associated text and visual appearance.A user initiates retrieval forsimilar trademarks by typing a text query.Subsequent searches can be performed by visual appearanceor using both appearance and text information.Textual information associated with trademarks issearched using the INQUERY search engine.Images are searched visually using a method for globalimage similarity by appearance developed in this paper.Images areﬁltered with Gaussian derivativesand geometric features are computed from theﬁltered images.The geometric features used here arecurvature and phase.Two images may be said to be similar if they have similar distributions of suchfeatures.Global similarity may,therefore,be deduced by comparing histograms of these features.This allows for rapid retrieval.The system’s performance on a database of2000trademark images isshown.A trademark database obtained from the US Patent and Trademark Ofﬁce containing63000design only trademark images and text is used to demonstrate scalability of the image search methodand multi-modal retrieval.1IntroductionRetrieval of similar trademarks is an interesting application for multimedia information retrieval.Con-sider the following example.The US Patent and Trademark Ofﬁce has a repository that has to be searchedfor conﬂicting(similar)trademarks before one can be awarded to a company or individual.There are several issues that make this task attractive for multi-modal information retrieval techniques.First,current searches are labour intensive.The number of trademarks stored is enormous and examiners have to leaf through large number of trademarks before making a decision.Second,there is a distinct notion of visual similarity used to compare trademarks.This is usually a decisive factor in an award decision.Third,there is readily available text information describing and categorizing a trademark.A system that automates these functions and helps the examiner decide faster would be immensely valuable.Clearly trademarks need to be searched both by text and image content.Text retrieval is a better understood problem,and there are several search engines that are applicable.However,the indexing and retrieval of images using their content is a difﬁcult problem.A person using an image retrieval system usually seeks toﬁnd semantically relevant information.For example,a person may be looking for a picture of a leopard from a certain viewpoint.Or alternatively,the user may require a picture of Abraham Lincoln from a particular viewpoint.Since the automatic segmentation of an image into objects is a difﬁcult and unsolved problem in computer vision,inferring semantic information from image content is difﬁcult to do.However,many image attributes like color,texture,shape and “appearance”are often directly correlated with the semantics of the problem.For example,logos or product packages(e.g.,a box of Tide)have the same color wherever they are found.The coat of a leopard has a unique texture while Abraham Lincoln’s appearance is uniquely deﬁned.These image attributes can often be used to index and retrieve images.In this paper,a system for multi-modal retrieval combining textual information and visual appearance is presented.The system combines text search using INQUERY[2]and image search.The image search was originally developed for general(heterogeneous)grey-level image collections[18].Here,it is applied to trademark images.Trademark images are large binary images rather than grey-level images.Trademark images may consist of geometric designs,more realistic pictures(for example,animals and people)as well as abstract images making them a challenging domain.Trademark images are also an example of a domain where there is an actual user need toﬁnd“similar”trademarks to avoid conﬂicts.Trademarks for this paper were obtained from the US Patent and Trademark ofﬁce.The63000design trademarks used here contain images of trademarks and associated text describing the trademark.Multi-modal retrieval begins with a user requesting trademarks that match a text query.The INQUERY search engine is used toﬁnd trademarks whose associated text match the query.The images associated with these trademarks are then displayed.Once an initial query is processed subsequent searches can be carried out by selecting the returned images and submitting them for retrieval by visual appearance or a combination of visual appearance and associated text.INQUERY is a well known search engine for retrieving text which is based on a probabilistic retrievalmodel called an inference net The reader is referred to[2]for details about the INQUERY engine.The current paper focuses on visual appearance representation,its quantitative evaluation with respect to trade-marks,scalability to a large collection,and feasibility to multi-modal retrieval.The visual appearance of an image is characterized here using the shape of the intensity surface.The images areﬁltered with Gaussian derivatives and geometric features are computed from theﬁltered im-ages.The geometric features used here are the image shape index(which is a ratio of curvatures of the three dimensional intensity surface)and the local orientation of the gradient.Two images are said to be similar if they have similar distributions of such features.The images are,therefore,ranked by compar-ing histograms of these features.Recall/Precision results with this method is tabulated with a database of about2000trademark images.Then multi-modal retrieval is demonstrated on a collection of63000 trademark images.The rest of the paper is organized as follows.Section2provides some background on the image retrieval area as well as on the appearance matching framework used in this paper.Section3surveys related work in the literature.In section4,the notion of appearance is developed further and characterized using Gaussian derivativeﬁlters and the derived global representation is discussed.Section5shows how the representation may be scaled for multi-modal retrieval from a database of about63,000trademark images.A discussion and conclusion follows in Section6.2Motivation and BackgroundThe different image attributes like color,texture,shape and appearance have all been used in a variety of systems for retrieving images similar to a query image(see3for a review).Systems like QBIC[6]and Virage[5]allow users to combine color,texture and shape to retrieve a database of general images.One weakness of such a system is that attributes like color do not have direct semantic correlates when applied to a database of general images.For example,say a picture of a red and green parrot is used to retrieve images based on their similarity in color with it.The retrievals may include other parrots and birds as well as redﬂowers with green stems and other images.While this is a reasonable result when viewed as a matching problem,clearly it is not a reasonable result for a retrieval system.The problem arises because color does not have a good correlation with semantics when used with general images.However,if the domain or set of images is restricted to sayﬂowers,then color has a direct semantic correlate and is useful for retrieval(see[3]for an example).Some attempts have been made to retrieve objects using their shape[6,22].For example,the QBIC system[6],developed by IBM,matches binary shapes.It requires that the database be segmented into objects.Since automatic segmentation is an unsolved problem,this requires the user to manually outline the objects in the database.Clearly this is not desirable or practical.Except for certain special domains,all methods based on shape are likely to have the same problem. An object’s appearance depends not only on its three dimensional shape,but also on the object’s albedo, the viewpoint from which it is imaged and a number of other factors.It is non-trivial to separate the different factors constituting an object’s appearance and it is usually not possible to separate an object’s three dimensional shape from the other factors.For example,the face of a person has a unique appearance that cannot just be characterized by the geometric shape of the’component parts’.In this paper a char-acterization of the shape of the intensity surface of imaged objects is used for retrieval.The experiments conducted show that retrieved objects have similar visual appearance,and henceforth an association is made between’appearance’and the shape of the intensity surface.Similarity can be computed using either local or global methods.In local similarity,a part of the query is used to match a part of a database image or images.One approach to computing local similarity[18]is to have the user outline the salient portions of the query(eg.the wheels of a car or the face of a person) and match the outlined portion of the query with parts of images in the database.Although,the technique works well in extracting relevant portions of objects embedded against backgrounds it is slow.The slow speed stems from the fact that the system must not only answer the question”is this image similar”but also the question”which part of the image is relevant”.This paper focuses on a representation for computing global similarity.That is,the task is toﬁnd images that,as a whole,appear visually similar.The utility of global similarity retrieval is evident,for example, inﬁnding similar scenes or similar faces in a face database.Global similarity also works well when the object in question constitutes a signiﬁcant portion of the image.2.1Appearance based retrievalThe image intensity surface is robustly characterized using features obtained from responses to multi-scale Gaussian derivativeﬁlters.Koenderink[14]and others[7]have argued that the local structure of an image can be represented by the outputs of a set of Gaussian derivativeﬁlters applied to an image.That is, images areﬁltered with Gaussian derivatives at several scales and the resulting response vector locally de-scribes the structure of the intensity surface.By computing features derived from the local response vector and accumulating them over the image,robust representations appropriate to querying images as a whole (global similarity)can be generated.One such representation uses histograms of features derived from the multi-scale Gaussian derivatives.Histograms form a global representation because they capture the distribution of local features(A histogram is one of the simplest ways of estimating a non parametric dis-tribution).This global representation can be efﬁciently used for global similarity retrieval by appearance and retrieval is very fast.The choice of features often determines how well the image retrieval system performs.Here,the task is to robustly characterize the3-dimensional intensity surface.A3-dimensional surface is uniquely de-termined if the local curvatures everywhere are known.Thus,it is appropriate that one of the features be local curvature.The principal curvatures of the intensity surface are invariant to image plane rotations, monotonic intensity variations and further,their ratios are in principle insensitive to scale variations of the entire image.However,spatial orientation information is lost when constructing histograms of curvature (or ratios thereof)alone.Therefore we augment the local curvature with local phase,and the representation uses histograms of local curvature and phase.Local principal curvatures and phase are computed at several scales from responses to multi-scale Gaus-sian derivativeﬁlters.Then histograms of the curvature ratios[13,4]and phase are generated.Thus,the image is represented by a single vector(multi-scale histograms).During run-time the user presents an example image as a query and the query histograms are compared with the ones stored,and the images are then ranked and displayed in order to the user.2.2The choice of domainThere are two issues in building a content based image retrieval system.Theﬁrst issue is technological, that is,the development of new techniques for searching images based on their content.The second issue is user or task related,in the sense of whether the system satisﬁes a user need.While a number of content based retrieval systems have been built([6,5]),it is unclear what the purpose of such systems is and whether people would actually search in the fashion described.In this paper we describe how the techniques described here may be scaled to retrieve images from a database of about63000trademark images provided by the US Patent and Trademark Ofﬁce.This database consists of all(at the time the database was provided)the registered trademarks in the United States which consist only of designs(i.e.there are no words in them).Trademark images are a good domain with which to test image retrieval.First,there is an existing user need:trademark examiners do have to check for trademark conﬂicts based on visual appearance.That is,at some stage they are required to look at the images and check whether the trademark is similar to an existing one.Second,trademark images may consist of simple geometric designs,pictures of animals or even complicated designs.Thus, they provide a test-bed for image retrieval algorithms.Third,there is text associated with every trademark and the associated text maybe used in a number of ways.One of the problems with many image retrieval systems is that it is unclear where the example or query image will come from.In this paper,the associated text is used to provide an example or query image.In addition associated text can also be combined with image ing trademark images does have some limitations.First,we are restricted to binary images(albeit large ones).As shown later in the paper,this does not create any problems for the algorithms described here.Second,in some cases the use of abstract images makes the task more difﬁcult.Others have attempted to get around it by restricting the trademark images to geometric designs[9].3Related WorkSeveral authors have tried to characterize the appearance of an object via a description of the intensity surface.In the context of object recognition[21]represent the appearance of an object using a parametric eigen space description.This space is constructed by treating the image as aﬁxed length vector,and then computing the principal components across the entire database.The images therefore have to be size and intensity normalized,segmented and trained.Similarly,using principal component representations described in[11]face recognition is performed in[26].In[24]the traditional eigen representation is augmented by using most discriminant features and is applied to image retrieval.The authors apply eigen representation to retrieval of several classes of objects.The issue,however,is that these classes are manually determined and training must be performed on each.The approach presented in this paper is different from all the above because eigen decompositions are not used at all to characterize appearance. Further,the method presented uses no learning and,does not require constant sized images.It should be noted that although learning signiﬁcantly helps in such applications as face recognition,however,it may not be feasible in many instances where sufﬁcient examples are not available.This system is designed to be applied to a wide class of images and there is no restriction per se.In earlier work we showed that local features computed using Gaussian derivativeﬁlters can be used for local similarity,i.e.to retrieve parts of images[18].Here we argue that global similarity can be determined by computing local features and comparing distributions of these features.This technique gives good results,and is reasonably tolerant to view variations.Schiele and Crowley[23]used such a technique for recognizing objects using grey-level images.Their technique used the outputs of Gaussian derivatives as local features.A multi-dimensional histogram of these local features is then computed.Two images are considered to be of the same object if they had similar histograms.The difference between this approach and the one presented by Schiele and Crowley is that here we use1D histograms(as opposed to multi-dimensional)and further use the principal curvatures as the primary feature.The use of Gaussian derivativeﬁlters to represent appearance is motivated by their use in describing the spatial structure[14]and its uniqueness in representing the scale space of a function[15,12,28, 25]The invariance properties of the principal curvatures are well documented in[7].Nastar[20],has independently used the image shape index to compute similarity between images.However,in his work curvatures were computed only at a single scale.This is insufﬁcient.In the context of global similarity retrieval it should be noted that representations using moment in-variants have been well studied[19].In these methods global representation of appearance may involve computing a few numbers over the entire image.Two images are then considered similar if these num-bers are close to each other(say using an L2norm).We argue that such representations are not able to really capture the“appearance”of an image,particularly in the context of trademark retrieval where mo-ment invariants are widely used.In other work[18]we compared moment invariants with the technique presented here and found that moment invariants work best for a single binary shape without holes in it, and,in general,fare worse than the method presented here.Jain and Vailaya[10]used edge angles and invariant moments to prune trademark collections and then use template matching toﬁnd similarity within the pruned set.Their database was limited to1100images.Texture based image retrieval is also related to the appearance based work presented in this ing Wold modeling,in[16]the authors try to classify the entire Brodatz texture and in[8]attempt to classify scenes,such as city and country.Of particular interest is work by[17]who use Gaborﬁlters to retrieve texture similar images.The earliest general image retrieval systems were designed by[6,22].In[6]the shape queries require prior manual segmentation of the database which is undesirable and not practical for most applications. 4Global representation of appearanceThree steps are involved in order to computing global similarity.First,local derivatives are computed at several scales.Second,derivative responses are combined to generate local features,namely,the principal curvatures and phase and,their histograms are generated.Third,the1D curvature and phase histograms generated at several scales are matched.These steps are described next.puting local derivatives:Computing derivatives usingﬁnite differences does not guarantee stability of derivatives.In order to compute derivatives stably,the image must be regularized,or smoothed or band-limited.A Gaussianﬁltered image obtained by convolving the image I with a normalized Gaussian is a band-limited function.Its high frequency components are eliminated and derivatives will be stable.In fact,it has been argued by Koenderink and van Doorn[14]and others [7]that the local structure of an image I at a given scale can be represented byﬁltering it with Gaussian derivativeﬁlters(in the sense of a Taylor expansion),and they term it the N-jet.However,the shape of the smoothed intensity surface depends on the scale at which it is observed.For example,at a small scale the texture of an ape’s coat will be visible.At a large enough scale,the ape’s coat will appear homogeneous.A description at just one scale is likely to give rise to many accidental mis-matches.Thus it is desirable to provide a description of the image over a number of scales,that is,a scale space description of the image.It has been shown by several authors[15,12,28,25,7],that under certain general constraints,the Gaussianﬁlter forms a unique choice for generating scale-space.Thus local spatial derivatives are computed at several scales.B.Feature Histograms:The normal and tangential curvatures of a3-D surface(X,Y,Intensity)are de-ﬁned as[7]:Where and are the local derivatives of Image I around point using Gaussian derivative at scale.Similarly,,and are the corresponding second derivatives.The normal curvature and tangential curvature are then combined[13]to generate a shape index as follows:when and is undeﬁned when either and are both zero,and is, therefore,not computed.This is interesting because veryﬂat portions of an image(or ones with constant ramp)are eliminated.For example in Figure1,the background in most of these images does not contribute to the curvature histogram.The curvature index or shape index is rescaled and shifted to the rangeas is done in[4].A histogram is then computed of the valid index values over an entire image.The second feature used is phase.The phase is simply deﬁned as. Note that is deﬁned only at those locations where is and ignored elsewhere.As with the curvature index is rescaled and shifted to lie between the interval.At different scales different local structures are observed and,therefore,multi-scale histograms are a more robust representation.Consequently,a feature vector is deﬁned for an image as the vectorwhere and are the curvature and phase histograms respectively.We found that using5scales gives good results and the scales are in steps of half an octave.C.Matching feature histograms:Two feature vectors are compared using normalized cross-covariance deﬁned aswhere.Retrieval is carried out as follows.A query image is selected and the query histogram vector is correlated with the database histogram vectors using the above formula.Then the images are ranked by their correlation score and displayed to the user.In this implementation,and for evaluation purposes,the ranks are computed in advance,since every query image is also a database image.4.1ExperimentsThe curvature-phase method is evaluated on a small database of2048images obtained from the US Patent and Trademark Ofﬁce(PTO).The images obtained from the PTO are large,binary and are converted to gray-level and reduced for the experiments.This smaller set is used because relevance judgments can be obtained relatively easily.In the following experiments an image is selected and submitted as a query.The objective of this query is stated and the relevant images are decided in advance.Then the retrieval instances are gauged against the stated objective.In general,objectives of the form’extract images similar in appearance to the query’will be posed to the retrieval algorithm.A measure of the performance of the retrieval engine can be obtained by examining the recall/precision table for several queries.Brieﬂy,recall is the proportion of the relevant material actually retrieved and precision is the proportion of retrieved material that is relevant[27].It is a standard widely used in the information retrieval community and is one that is adopted here.Figure1:Trademark retrieval using Curvature and PhaseQueries were submitted for the purpose of computing recall/precision.The judgment of relevance is qualitative.For each query in both databases the relevant images were decided in advance.These were restricted to48.The top48ranks were then examined to check the proportion of retrieved images that were relevant.All images not retrieved within48were assigned a rank equal to the size of the database.Table1:Precision at standard recall points for six QueriesRecall1030507090 Precision(trademark)%93.285.274.545.59.0Precision(assorted)%92.688.386.865.912.061.1%66.3%That is,they are not considered retrieved.These ranks were used to interpolate and extrapolate precision at all recall points.In the case of assorted images relevance is easier to determine and more similar for different people.However in the trademark case it can be quite difﬁcult and therefore the recall-precision can be subject to some error.The recall/precision results are summarized in Table1and both databases are individually discussed below.Figure1shows the performance of the algorithm on the trademark images.Each strip depicts the top 8retrievals,given the leftmost as the query.Most of the shapes have roughly the same structure as the query.Note that,outline and solidﬁgures are treated similarly(see rows one and two in Figure1).Six queries were submitted for the purpose of computing recall-precision in Table1.Tests were also carried out with an assorted collection of1561grey-level images.These results are discussed elsewhere[1],and the recall/precision table is shown in Table1.While the queries presented here are not“optimal”with respect to the design constraints of global similarity retrieval,they are however,realistic queries that can be posed to the system.Mismatches can and do occur.Theﬁrst is the case where the global appearance is very different.Second,mismatches can occur at the algorithmic level.Histograms coarsely represent spatial information and therefore will admit images with non-trivial deformations.The recall/precision presented here compares well with text retrieval.The time per retrieval is of the order of milli-seconds.In the next section we discuss the application of the presented technique to a database of63000images.5Trademark RetrievalThe system indexes63,718trademarks from the US Patent and Trademark ofﬁce in the design only category.These trademarks are binary images.In addition,associated text consists of a design code that designates the type of trademark,the goods and services associated with the trademark,a serial number and a short descriptive text.The system for browsing and retrieving trademarks is illustrated in Figure2.The netscape/Java user interface has two search-able parts.On the left a panel is included to initiate search using text.Any or all of theﬁelds can be used to enter a query.In this example,the text“Merriam Webster’is entered.All images associated with it are retrieved using the INQUERY[2]text search engine.The user can then use any of the example pictures to search for images that are similar visually or restrict it to images withTable2:Fields supporting the text query FieldThe business this trademark is used inAll are of type DESIGN ONLYAn assigned code categorySerial number assigned to trademarkDate trademark application wasﬁledNumber assigned to trademarkDate trademark was registeredOwner of the trademarkA textual description of the trademark. Section44Type of markRegisterAfﬁdavit textLive/dead(1,4,8).A histogram descriptor of the image is obtained by concatenating all the individual histograms across scales and regions.These two steps are conducted off-line.Execution:The image search server begins by loading all the histograms into memory.Then it waits on a port for a query.A CGI client transmits the query to the server.Its histograms are matched with the ones in the database.The match scores are ranked and the top requested retrievals are returned.5.1ExamplesFigure2:Retrieval in response to a“Merriam Webster”queryIn Figure2,the user typed in Merriam Webster in the text window.The system searches for trade-marks which have either Merriam or Webster in th associated text and displays them.Here,theﬁrst two trademarks(ﬁrst two images in the left window)belong to Merriam Webster.In this example,the user has chosen to’click’the second image and search for images of similar trademarks.This search is based entirely on the image and the results are displayed in the right window in rank order.Retrieval takes a few seconds and is done by comparing histograms of all63,718trademarks on theﬂy.。

传统文化赋能新能源汽车品牌战略初探

70AUTO TIMENEW ENERGY AUTOMOBILE | 新能源汽车作为第二次工业革命与现代化文明的明珠，汽车从20世纪初诞生至今，已经走过了一百多年的漫漫历程，如今已成为人们日常出行中不可或缺的交通工具。

传统汽车多以汽油、柴油为能量来源，依赖不可再生能源，但当前在地缘冲突、气候变化、能源危机等多种因素的影响下，全球能源安全不确定性加剧，摆脱对于不可再生资源的依赖，寻求新的能源已然成为迫在眉睫的问题。

同时，党的二十大报告指出，我国力争2030年前二氧化碳排放达到峰值，努力争取2060年前实现碳中和目标。

能源安全问题和双碳目标的设置，迫切需要我国能源结构加快向绿色低碳、节能环保转型，构建清洁低碳、安全高效的新型能源体系。

在此愿景之下，新能源汽车应运而生，与传统的汽车相比，新能源汽车的主要能量来源是动力电池，动力电池提供电能来驱动车辆行驶，由于电能的来源多样化，使得能够摆脱对于不可再生能源石油的依赖；同时，新能源汽车行驶过程中无有害气体排放，对于环境更加友好。

近十年来，随着政府对新能源汽车的政策支持，新能源汽车行业的产业升级，新能源汽车企业的技术创新等因素，新能源汽车的发展迎来了重大机遇，中国新能源汽车的发展实现弯道超车，展现出了强大的创新力王芳芳　于蕾烟台汽车工程职业学院　山东省烟台市　265500摘要：能源安全问题和双碳目标的设置，迫切需要我国能源结构加快向绿色低碳、节能环保转型，构建清洁低碳、安全高效的新型能源体系。

在此愿景之下，新能源汽车迎来重大机遇，实现弯道超车。

国产新能源汽车的超越式发展，与愈发“坚硬”的文化“软实力”有着密切关联。

而提升国产新能源汽车的文化内涵，向海外大量输入具有中国特色、中国文化气派的工业化汽车产品，在提升工业产品影响力的同时，对于文化的传播、话语权的建立、民众对于中国制造与中国文化的欣赏与信赖具有重大意义。

关键词：新能源汽车　传统文化　软实力　中国声音传统文化赋能新能源汽车品牌战略初探和竞争力。

Owner’s Manual For use with models 1000G3, 3000G3

Owner’s ManualThis Manual Includes:• Warranty• Safety Warnings• Operating Instructions • Replacement Directions • Troubleshooting Tips• Cleaning and Maintenance TipsFor use with models:1000G3, 3000G3, 3000XG3Before operating this unit carefullyread the contents of this manual.Read all instructions carefully before operatingyour air purifier.Table of ContentsSafety Instructions and Cautions2 Warranty3 Maintenance and Customer Service4 Placement and Operation4 Cleaning4 AHPCO® Cell Replacement5-6 Troubleshooting Guide7WARNING: UV Light Hazard. Harmful to skin and eyes. Can cause temporary or permanent loss of vision. Never look at the lamp while illuminated. To prevent exposure to ultraviolet light, be sure the pow-er is disconnected before servicing.WARNING: RISK OF ELECTRICAL SHOCK. CAN CAUSE INJU-RY OR DEATH: UNPLUG OR DISCONNECT UNIT FROM POWER SUPPLY BEFORE SERVICING.Replacement Date ReminderThe AHPCO ® Cell should be replaced at least every three years.Replacement Date:_________________Hg lamp contains mercury. Manage in accord with disposal laws. Find disposal centers at .Important Safety InstructionsWhen operating electrical appliances, basic precautions should always be fol-lowed.WARNING To reduce the risk of fire, electric shock or injury:• Do not use outdoors or on wet surfaces.• Use only as described in this manual.• Do not use with damaged cord or plug. If appliance is not working as it should, has been dropped, damaged, left outdoors, or dropped into water, call customer service at 1-800-936-1764.• Do not unplug by pulling on cord. To unplug, grasp the plug, not the cord.• Do not handle plug or appliance with wet hands.• Do not put any object into openings.• Turn off all controls before unplugging.AHPCO ® Cell contains Hg (Mercury) and should be disposed of according to dis-posal laws. Find disposal centers at .If the UV lamp is broken do not touch the cell or glass with your hands.UV lamp may be hot and could cause serious burns if not handled properly. Please wait until the AHPCO ® Cell has cooled to room temperature to remove from unit.CAUTIONWarrantyAir Oasis gives you the following limited warranty for this product only if it was originally purchased directly from Air Oasis or an Air Oasis Authorized Retail Dealer.Air Oasis will repair or replace, free of charge, to the original purchaser, any part that is found to be defective in material or workmanship within three (3) years of the date of purchase.This limited warranty covers the replacement of expendable or consumable parts such as the AHPCO® Cell for two (2) years.This limited warranty does not apply to any part subjected to accident, abuse, industrial use, alteration, misuse, damage caused by act of God, the use of voltages other than indicated on the label displayed on this product or service of this product by anyone other than Air Oasis.Air Oasis does not authorize any person or representative to assume or grant any other warranty obligation with the sale of this product.Air Oasis’ limited warranty is valid only if you retain proof of purchase from Air Oasis or an Air Oasis Authorized Retail Dealer for this product. If you purchase this product from any other source, your purchase is “AS IS”, which means Air Oasis grants you no warranty, and that you, not Air Oasis, assume the entire risk of the quality and performance of this product, including the entire cost of any necessary servicing or repairs of any defects.Air Oasis’ liability for damages to you for any costs whatsoever arising out of this statement of limited warranty shall be limited to the amount paid for this product at the time of original purchase, and Air Oasis shall not be liable for any direct, indirect, consequential or incidental damages arising out of the use or inability to use this product.ALL EXPRESS AND IMPLIED WARRANTIES FOR THIS PRODUCT, INCLUDING IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, ARE LIMITED IN DURATION TO THE WARRANTY PERIOD, AND NO WARRANTIES, WHETHER EXPRESS OR IMPLIED, WILL APPLY AFTER THIS PERIOD.Some states do not allow limitations on the duration of implied warranties, so the above limitation may not apply to you. This warranty gives you specific legal rights, and you may also have other rights which vary from state to state.To obtain warranty service and return authorization number goto /rma and fill out the form provided.NOTE: All returned packages that do not have an RMA# will be refused.Maintenance and Customer ServicePlacement and OperationOptional CleaningIf you require any additional information or have problems with your Air Oasis ap-pliance, you may call Air Oasis customer service at:1-800-936-1764Please have your serial and model numbers, found on bottom of unit, handy when calling.Save your sales receipt to show if your Air Oasis appliance should ever need any warranty service.Off / On SwitchBase and Honeycomb - Werecommend wiping with a dry towel toremove dust and build up.Aluminum Shell - We recommend using stainless steel/aluminum spray on polish or multi-surface cleaners. Make sure the polish or cleaner is recommended for use on aluminum.Internal - Take dry cloth and wipe inside the shell and around the aluminum brackets and the ballast or spray with compressed air.NOTE: Use compressed air only to clean AHPCO Cell. Any removal of catalyst coating will decrease the effectiveness of your unit and void the warranty.Bi-Polar Brush Heads - The Bi-Polar brush heads inside your unit should be cleaned periodically to remove any debris for optimal effectiveness.For best results the unit should be placed in a central location and three feet or higher from the ground.Operation is very simple. Remove the Air Oasis from the packaging.Place in desired location, plug it in and rock the switch to the on position. The unit can remain in the on position 24/7.Upon startup you will hear 3 beeps signifying the unit has power. After a brief delay of less than 4 seconds the AHPCO ® Cell will illuminate and you will see a blue glow inside the unit.Make sure that the power cord is unplugged prior to cleaning.AHPCO™ Cell ReplacementStep 1CAUTION: Before attempting to service the Air Oasis unit, be sure the power is off and unplugged.DANGER - NEVER LOOK DIRECTLY AT ILLUMINATED AHPCO CELL1. Remove two (2) screws at the bottom sides holding outer shell in place. Then slide the shell up and off of the internal base.Step 2Step 33. Grasp the AHPCO ® Cell and pull to the side, then remove the lamp connector.2. Use a phillips headscrewdriver to remove the screws from the AHPCO ® Cell.CAUTIONUV lamp may be hot and could cause serious burns if not handled properly. Please wait until AHPCO ® Cell has cooled to room temperature to removeWARNINGOPTICAL RADIATION EXPOSURE HAZARDDO NOT ATTEMPT TO OPER ATE UNIT WITHOUTALLUMINUM SHELL COVERING AHPCO CELL.DO NOT ATTEMPT TO REPLACEAHPCOCELL WITHOUT DISCONNECTING POWER.PERMANENT EYE AND SKIN DAMAGE MAY RESULT .Step 4Step 5Step 65. Connect the lamp connector securely onto the AHPCO ® Cell and use a phillips headscrewdriver to tighten the screws to secure the AHPCO ® Cell.6. Slide shell down onto base and replace screws. (Cord and switch are to the back.)4. If unit was purchased after Nov. 2016, before you plug in the AHPCO ® Cell, you will need to plug in the power cord, turn the unit on and press and hold the reset button for 3 seconds to make it stop beeping when replacing the AHPCO ®Cell.Troubleshooting GuideProblemReasonSolutionUnit does not turn on.1. Power Supply2. Switch3. AHPCO ® Cell1. Check to make sure power supply is plugged in.2. Check to make sure switch is in the on position.3. Check to make sure UV lamp within AHPCO ® Cell in plugged in completely.Single Beep Repeated Slowly.1. The AHPCO ® Cell has Expired. 1.Replace the AHPCO ® Cell.3 Beeps Repeated1. The AHPCO ® Cell is not plugged in.2. The UV Lamp is defective.1. Remove outer shell to check that AHPCO ® Cell is properly plugged in.2. Replace the AHPCO ® Cell.3401 Airway Blvd.Amarillo, TX 79118Toll-Free: 1-800-936-1764 Fax: 1-806-373-7799Purifier doesn’t seem to be working to optimal efficiency.1. Purifier is being operated in areas with a highconcentration of particulate matter.1. Refer to AHPCO Cell replacementInstructions to remove shell. Use compressed air or vacuum toremove dust & debris.。

AXIS P13网络摄像头系列说明书

FICHE TECHNIQUELa série de caméras réseau AXIS P13 comporte des caméras fixes pour l’intérieur et pour l’extérieur qui offrent une qualité d’image exceptionnelle avec une compression H.264. Ces caméras sont idéales pour une surveillance hautes performances. Les modèles mégapixel offrent également une video HDTV 720p/1080p.Série de caméras réseau AXIS P13Superbe qualité d’image pour la vidéosurveillance dans tous types d’environnements.> Superbe qualité vidéo incluant la HDTV et 5 megapixels > Contrôle P-Iris > Multiples flux vidéo H.264> PTZ numérique et flux de vues multiples > Modèles utilisables à l’extérieurLa série des AXIS P13 offre des caméras supportant toute éten-due de résolutions allant jusqu’à 5 mégapixels, notamment avec les caméras AXIS P1347 et AXIS P1347-E. Les modèles sont dis-ponibles à la fois dans les versions intérieur et extérieur ”-E ”. Les caméras fournissent une large gamme dynamique, une fonc-tionnalité jour et nuit avec une superbe qualité d’images dans des conditions de luminosité comme d’obscurité.Les caméras 3- et 5-megapixel proposent également un contrôle P-I ris unique et révolutionnaire, qui leur permet de contrôler avec précision la position de l’iris afin d’optimiser la profondeur du champ et la résolution de l’objectif pour obtenir une qualité d’image optimale.Toutes les caméras AXIS P13 prennent en charge plusieurs flux de données vidéo au format H.264 et Motion JPEG. La techno-logie H.264 réduit de manière significative les exigences en bande passante et en stockage sans affecter la qualité de l’image.Les modèles SVGA et mégapixels ont une fonction de mise au point arrière à distance qui permet à la mise au point d’êtreajustée depuis un ordinateur. Ces mêmes modèles offrent égale-ment une fonction panoramique/inclinaison/zoom numérique, tandis que les caméras AXI S P1346/-E sont en plus équipées d’un flux à vues multiples. Les caméras AXI S P13 prennent en charge l’alimentation par Ethernet (PoE), ce qui facilite leur installation. Les modèles ex-térieurs sont alimentés par Ethernet et par High PoE et fonc-tionnent à des températures comprises entre -40 ºC et 50 ºC .La série AXIS P13 comprend des caméras réseau fixes conçues pour l’intérieur et l’extérieur qui conviennent à un large éventail d’applications de vidéosurveillance, bâtiments publics ou industriels, commerces, aéroports, gares et écoles. Caméras hautes performances pour l’intérieur/l’extérieurInstallation facile grâce à l’assistance à la mise au point, la mise au point à distance et le compteur de pixelsL’assistance à la mise au point simplifie le réglage de la mise au point de toutes les caméras AXIS P13, grâce au clignotement d’un voyant vert lorsqu’une image est au point après un réglage manuel de l’objectif. De plus, les modèles SVGA et mégapixel/HDTV sont équipés d’une fonction de mise au point à distance du foyer arrière qui permet l’ajustement de la mise au point à partir d’un ordina-teur. Le compteur de pixels permet de vérifier que la résolution en pixels d’un objet est conforme aux règlementations en vigueur ou aux besoins du client (ex. : reconnaissance faciale).Modèles utilisables àl’extérieurLes caméras réseau AXI S P13-E permet-tent de gagner du temps et de réaliser deséconomies puisqu’elles sont immédiate-ment prêtes pour un montage en exté-rieur. Certifiées IP66, elles sont protégéescontre la poussière, la pluie, la neige et lesrayons du soleil et peuvent fonctionnerjusqu’à une température de -40 °C. Lescaméras sont alimentées par Ethernet, cequi facilite l’installation puisqu’elles nenécessitent pas de câble d’alimentation séparé. Une membrane de déshumidifi-cation intégrée permet d’éliminer toute l’humidité le boîtier de la caméra lors de l’installation. Ces caméras permettent l’installation facile d’une lampe à infrarouge sous le boîtier. Ils arrivent avec un support de montage mural, un pare-soleil et les presse-étoupes.PTZ numérique et flux à vue multipleLes modèles de caméras SVGA et mégapixel sont dotés de fonctions pan-oramiques, d’inclinaison et de zoom numériques qui permettent de sélectionnerune vue détourée de la vue d’ensemble pour l’afficher ou l’enregistrer, ce quiréduit ainsi le débit binaire et l’espace de stockage requis. Les caméras 3- et5-megapixel sont également dotées de la fonction de flux à vues multiplesqui permet de transmettre simultanément plusieurs zones détourées de la vuecomplète, simulant jusqu’à huit caméras virtuelles.Flux à vues multiples avec les cameras réseau AXIS P1346/-E et AXIS P1347/-EUne caméra Vue panoramique complète offrant des zonesdétourées de la vue complètePlusieurs champs de vision virtuels dela caméra (jusqu’à huit vues possibles)Contrôle P-IrisLes cameras 3-megapixel AXI S P1346/-E et 5-megapixel AXI S P1347/-E of-frent un nouveau contrôle précis de l’iris avancé, P-Iris, qui établit de nouvelles normes de qualité d’image pour les caméras fixes. Ce contrôle comporte un objectif P-Iris spécial associé à un logiciel spécialisé de la caméra qui fournit la meilleure position de l’iris pour un contraste, une clarté, une résolution et une profondeur de champ améliorés de l’image. Une bonne profondeur de champ, où des objets situés à différentes distances de la caméra sont simultanément mis au point, permet d’obtenir une meilleure visibilité de scène.Le P-Iris est particulièrement utile aux caméras mégapixel/HDTV, car il permet de continuer à obtenir des images haute résolution nettes, même dans des conditions d’éclairage difficiles. Il utilise le même type de connecteur et de câble que l’iris DC classique qui est également pris en charge par les caméras 3- et 5-megapixel pour la rétrocompatibilité.Pour en savoir plus sur P-Iris et ses contrôles, cliquez sur le lien :/corporate/corp/tech_papers.htmAXIS P1343/P1344/P1346/P1347 :Microphone intégréPour plus d’informations, visitez le site ** Ce produit inclut un logiciel développé par le projet OpenSSLpour une utilisation dans la boîte à outils OpenSSL. ()©2012 Axis Communications AB. AXIS COMMUNICATIONS, AXIS, ETRAX, ARTPEC et VAPIX sont des marques déposées d’Axis AB ou en cours de dépôt par Axis AB dans différentes juridictions. Tous les autres noms, produits ou services sont la propriété de leurs détenteurs respectifs. Document sujet à modification sans préavis.A x i s C o m m u n i c a t i o n s S A S -R C S B 4 0 8 9 6 9 9 9 8 4 7 2 9 2 / F R / R 1 / 1 2 0 4AXIS T8123 à 1 port。

KSZ8842-16MQL MVL 2-Port 交换机和通用总线接口应用注释132：语音、视频和数

Application Note 132Voice Video and Data Communications using a2-Port Switch and Generic Bus InterfaceKSZ8842-16MQL/MVLIntroductionThe IP-Telephony market is booming, due to the ease of use of the technology and the low cost it promises consumers for making voice and video calls using the technology. Conveniently, it runs on the ever popular Ethernet LAN technology, which currently supports over 95 percent of all companies’ networks, and works like computer data on the LAN. Using VoIP (Voice over Internet Protocol), voice signals can be packetized like computer data packets and transmitted worldwide over the Internet. This is why VoIP is a win-win solution for everyone.Packet-switched VoIP converts voice signals into packets and each VoIP packet includes both the sender’s/receiver’s network addresses. The VoIP packets can traverse any IP network and choose alternate paths because the destination address is included in the packet, so the VoIP is flexible, interoperable and portable.IP telephone has a NID (Network Interface Device) built into it just like a computer must have a NIC (Network Interface Card) inside of it to connect to the LAN. The NID is the single most important component for VoIP applications and LAN connections because it provides its physical address on the LAN.Micrel’s KSZ8842-16MQL, a network interface device, is the industry’s first 2-port Ethernet Layer-2 Managed switch and 8/16 bit generic host bus interface as well as:A KSZ8842-32MQL, 32-bit Generic Bus (synchronous or asynchronous) for different host processor interfaces1K Dynamic and 8 Static MAC addresses, 16 VLAN entries and 34 MIB Counters per portSupports programmable rate limiting on the ingress/egress ports and 4 priority queuesDynamic Packet Memory scheme which fully utilizes 8KB buffer memory in 4B incrementsLinkMD cable diagnostic capabilities. Determines distance to fault and also cable length measurementPrevents congestion in the switch with a MAC filtering function to filter or forward unknown unicast packets (useful for VoIP applications)HP Auto-MDIX crossover with disable and enable optionSeparate Link and Activity pins on all portsAlternative LQFP packaging available in the KSZ8842-16MVL and -32MVL devicesFigure 1 below shows a system level block diagram of the KSZ8842-16M being used in a VoIP application.To/FrSwitchIP PhoneFigure 1: KSZ8842-16MQL Embedded Two-port Ethernet Switch for VoIP ApplicationsA detailed VoIP phone block diagram is shown in Figure 2 below. The KSZ8842M is designed to handle voice, video and data packets between the LAN connection and Host CPU interface. The following sections focus on how to connect the KSZ8842-16MQL 16-bit generic bus interface to a DSP Host processor.LAN-Port1LAN-Port2Figure 2: VoIP Phone system level Block DiagramGeneral Description the Bus Interface between KSZ8842-16M and ProcessorThe KSZ8842-16MQL/MVL is 2-Port Ethernet Switch with non-PCI interface, and is designed to connect to an 8 or 16-bit bus interface. This application note provides a basic overview of system level signal connections based on 8 or 16-bit bus interfaces, in combination with or without EEPROM.The KSZ8842-16MQL/MVL supports two transfer modes in the BIU (Bus Interface Unit):Asynchronous modeSynchronous modeBoth asynchronous and synchronous signals are independent of each other.In order to handshake in the different bus interfaces (ISA-like, EISA-like or VLBus-like), the following sections will describe all bus interface signal connections using these two transfer modes.8 or 16-Bit Bus Interface Signal Connections for the KSZ8842-16MQL/MVL8 or 16-Bit Asynchronous Bus Interface ModeIn the asynchronous bus interface mode, the KSZ8842-16MQL/MVL host bus read/write operation is as an 8 or 16-bit peripheral. All signals are listed in Table 1 and connections are shown in Figures 3, 4, and 5 respectively for 8-bit configurations, and in Figures 6, and 7 for 16-bit operations. The timing waveform is shown in Figure 8.Table 1: KSZ8842-16MQL/MVL Bus Interface Signals for 8 or 16-Bit Asynchronous ModeAsynchronousSignal TypeADSN = 0 (ISA-like) Using ADSN (EISA-like) A[15:1] IAddress AddressD[15:0] I/O Data (8 or 16-bit) Data (8 or 16-bit)AEN I Address Enable (active low) Address Enable (active low)BE1N,BE0N I Byte Enable (active low) Byte Enable (active low)ADSN I Always enabled Address Strobe (Tied low) Address Strobe is used to latch A[15:1], AEN,BE1N/BE0N LDEVN O Local Device (asserted low when right address decoded) Local Device (asserted low when right address decoded)INTRN O Interrupt (asserted low when interrupt status bit set) Interrupt (asserted low when interrupt status bit set)RDN I Asynchronous Read (active low) Asynchronous Read (active low)WRN I Asynchronous Write (active low) Asynchronous Write (active low)ARDY O Asynchronous Ready (active high) Asynchronous Ready (active high)VLBUSN I Tied high for non-VLBus Tied high for non-VLBusCYCLEN I Not used (Tied high) Not used (Tied high)SWR I Not used (Tied high) Not used (Tied high)RDYRTNN I Not used (Tied high) Not used (Tied high)BCLK I Not used (Tied low) Not used (Tied low)SRDYN O Not used (No connect) Not used (No connect)Note: These signals BE3N, BE2N, DATACSN are not available (No Connect) in KSZ8842-16MQL/MVL deviceFigure 3: 8-Bit Asynchronous ISA-like Bus Connections with A[3:1]Figure 4: 8-Bit Asynchronous ISA-like Bus Connections with A[15:1]Figure 5: 8-Bit Asynchronous ISA-like Bus Connections with EEPROMFigure 6: 16-Bit Asynchronous ISA-like Bus Connections without EEPROMFigure 7: 16-Bit Asynchronous ISA-like Bus Connections with EEPROMAddr, AEN, BExNADSN Read Data RDN, WRN Write Data ARDY (Read Cycle) ARDY ( W rite Cycle)Figure 8: Asynchronous Read & Write Cycles Timing Waveform – ADSN = 0Symbol Parameter Min Typ Max Unitt1 A1-A15, AEN, BExN[3:0] valid to RDN, WRN active 2 ns1 nst2 A1-A15, AEN, BExN[3:0] hold after RDN, WRNinactive (assume ADSN tied Low)t3 Read data valid to ARDY rising 0.8 nst4 Read data to hold RDN inactive 4 nst5 Write data setup to WRN inactive 4 nst6 Write data hold after WRN inactive 2 nst7 Read active to ARDY Low 8 nst8 Write inactive to ARDY Low 8 ns0 110 nst9 ARDY low (wait time) in read cycle (Note1)(It is 0ns to read bank select register)(It is 110ns to read QMU data register)0 85 nst10 ARDY low (wait time) in write cycle (Note1)(It is 0ns to write bank select register)(It is 85ns to write QMU data register)Note1: When CPU finished current Read or Write operation, it can do next Read or Write operation even theARDY is low. During Read or Write operation if the ADRY is low, the CPU has to keep the RDN/WRN low until theARDY returns to high.16-Bit Synchronous Bus Interface ModeIn the synchronous bus interface mode, the KSZ8842-16MQL/MVL host bus read/write operation is as a 16-bit peripheral. All signals are listed in the Table 2 and connections are shown in Figures 9 and 10. The timing waveform is shown in Figures 11 and 12.Table 2: KSZ8842-16MQL/MVL Bus Interface Signals for 16-Bit Synchronous ModeSynchronousSignal TypeVLBUSN = 0 (VLBus-like)AddressA[15:1] ID[15:0] I/O Data (8 or 16-bit)AEN I Address Enable (active low)BE1N, BE0N I Byte Enable (active low)ADSN I Address Strobe is used to latch A[15:1], AEN,BE1N/BE0NLDEVN O Local Device (asserted low when right address decoded)INTRN O Interrupt (asserted low when interrupt status bit set)RDN I Not used (Tied high)WRN I Not used (Tied high)NotusedARDY OVLBUSN I Tied low for VLBus-like cycleCYCLEN I CYCLEN is used to sample SWR when it is assertedSWR I Synchronous write cycles when high and read cycles when lowRDYRTNN I Ready Return is used by the Host to indicate the end of Read or Write in VLBus-like cycleBCLK I Bus Clock is used for Synchronous transferSRDYN O Synchronous Ready is used to indicate that data is ready to Read/WriteFigure 9: 16-Bit Synchronous VLBUS-like Bus Connections without EEPROMFigure 10: 16-Bit Synchronous VLBUS-like Bus Connections with EEPROMFigure 11: Synchronous Write Cycle Timing Waveform – VLBUS = 0Symbol Parameter Min Typ Max Unit t1 A1-A15, AEN, BExN[3:0] setup to ADSN rising 4 nst2 A1-A15, AEN, BExN[3:0] hold after ADSN rising 2 nst3 CYCLEN setup to BCLK rising 4 nst4 CYCLEN hold after BCLK rising (non-burst mode) 2 nst5 SWR setup to BCLK 4 nst6 SWR hold after BCLK rising with SRDYN active 0 nst7 Write data setup to BCLK rising 5 nst8 Write data hold from BCLK rising 1 nst9 SRDYN setup to BCLK 8 nst10 SRDYN hold to BCLK 1 nst11 RDYRTNN setup to BCLK 4 nst12 RDYRTNN hold to BCLK 1 nsBCLKAddress, AEN, BExNADSNSWRCYCLENRead DataSRDYNRDYRTNNFigure 12: Synchronous Read Cycle Timing Waveform – VLBUS = 0Symbol Parameter Min Typ Max Unit t1 A1-A15, AEN, BExN[3:0] setup to ADSN rising 4 ns t2 A1-A15, AEN, BExN[3:0] hold after ADSN rising 2 ns t3 CYCLEN setup to BCLK rising 4 ns t4 CYCLEN hold after BCLK rising (non-burst mode) 2 ns t5 SWR setup to BCLK 4 ns t6 Read data hold from BCLK rising 1 ns t7 Read data setup to BCLK 8 ns t8 SRDYN setup to BCLK 8 ns t9 SRDYN hold to BCLK 1 ns t10 RDYRTNN setup to BCLK rising 4 ns t11 RDYRTNN hold after BCLK rising 1 nsMicrel Confidential KS8842-16MQL/MVL AN 132ConclusionBy using this Application Note, customers are able to design a VoIP phone system with the KSZ8842-16MQL/MVL to easily connect to their FPGAs or processors as well as any other application requiring a two-port switch and generic bus interface for Embedded and Industrial Ethernet applications.In addition, Micrel provides the flexibility of offering a single port KSZ8841-16MQL/MVL MAC/PHY plus generic bus interface part that is 100% footprint compatible for single port applications. This provides engineers with the flexibility to design two products using a single print circuit board and software driver, thereby saving time, money and efforts in the development cycle.All of the development collateral including data sheet, schematics, gerber file, IBIS module and software driver can be downloaded from Micrel website. Evaluation boards and user’s guide are also available.MICREL, INC. 1849 FORTUNE DRIVE SAN JOSE, CA 95131 USATEL +1 (408) 944-0800 FAX +1 (408) 474-1000 WEB http:/The information furnished by Micrel in this data sheet is believed to be accurate and reliable. However, no responsibility is assumed by Micrel for its use.Micrel reserves the right to change circuitry and specifications at any time without notification to the customer.Micrel Products are not designed or authorized for use as components in life support appliances, devices or systems where malfunction of a product can reasonably be expected to result in personal injury. Life support devices or systems are devices or systems that (a) are intended for surgical implant into the body or (b) support or sustain life, and whose failure to perform can be reasonably expected to result in a significant injury to the user. A Purchaser’s use or sale of Micrel Products for use in life support appliances, devices or systems is a Purchaser’s own risk and Purchaser agrees to fully indemnifyMicrel for any damages resulting from such use or sale.© 2005 Micrel, Incorporated.December 2005 11 Rev. 1.0。

Review of maglev train technologies

Review of Maglev Train TechnologiesHyung-Woo Lee1,Ki-Chan Kim2,and Ju Lee2Korea Railroad Research Institute,Uiwang437-757,KoreaDepartment of Electrical Engineering,Hanyang University,Seoul133-791,KoreaThis paper reviews and summarizes Maglev train technologies from an electrical engineering point of view and assimilates the results of works over the past three decades carried out all over the world.Many researches and developments concerning the Maglev train have been accomplished;however,they are not always easy to understand.The purpose of this paper is to make the Maglev train technologies clear at a glance.Included are general understandings,technologies,and worldwide practical projects.Further research needs are also addressed.Index Terms—EDS,EMS,Maglev train,magnetic guidance,magnetic levitation,magnetic propulsion.I.I NTRODUCTIONA LONG with the increase of population and expansionin living zones,automobiles and air services cannot afford mass transit anymore.Accordingly,demands for in-novative means of public transportation have increased.In order to appropriately serve the public,such a new-generation transportation system must meet certain requirements such as rapidity,reliability,and safety.In addition,it should be convenient,environment-friendly,low maintenance,compact, light-weight,unattained,and suited to mass-transportation.The magnetic levitation(Maglev)train is one of the best candidates to satisfy those requirements.While a conventional train drives forward by using friction between wheels and rails,the Maglev train replaces wheels by electromagnets and levitates on the guideway,producing propulsion force electromechanically without any contact.The Maglev train can be reasonably dated from1934when Hermann Kemper of Germany patented it.Over the past few decades since then,development of the Maglev train went through the quickening period of the1960s,the maturity of the1970s–1980s,and the test period of the1990s,ﬁnally accomplishing practical public service in2003in Shanghai, China[1]–[4].Since the Maglev train looks to be a very promising solu-tion for the near future,many researchers have developed tech-nologies such as the modeling and analysis of linear electric machinery,superconductivity,permanent magnets,and so on [5]–[25].The Maglev train offers numerous advantages over the con-ventional wheel-on-rail system:1)elimination of wheel and track wear providing a consequent reduction in maintenance costs[26];2)distributed weight-load reduces the construction costs of the guideway;3)owing to its guideway,a Maglev train will never be derailed[96];4)the absence of wheels removes much noise and vibration;5)noncontact system prevents it from slipping and sliding in operation;6)achieves higher grades and curves in a smaller radius;7)accomplishes acceleration and de-celeration quickly;8)makes it possible to eliminate gear,cou-pling,axles,bearings,and so on;9)it is less susceptible toDigital Object Identiﬁer10.1109/TMAG.2006.875842TABLE IC OMPARISON OF M AGLEV AND W HEEL-ON-R AIL SYSTEMS weather conditions.However,because there is no contact be-tween rails and wheels in the Maglev train,the traction mo-tors must provide not only propulsion but also braking forces by direct electromagnetic interaction with the rails.Secondly, the more weight,the more electric power is required to support the levitation force,and it is not suitable for freight.Thirdly, owing to the structure of the guideway,switching or branching off is currently difﬁcult.Fourthly,it cannot be overlooked that the magneticﬁeld generated from the strong electromagnets for levitation and propulsion has effects on the passenger compart-ment.Without proper magnetic shielding,the magneticﬁeld in the passenger compartment will reach0.09T atﬂoor level and 0.04T at seat level.Suchﬁelds are probably not harmful to human beings,but they may cause a certain amount of inconve-nience.Shielding for passenger protection can be accomplished in several ways such as by putting iron between them,using the Halbach magnet array that has a self-shielding characteristic, and so on.[27],[79].Table I shows the comparison of Maglev and wheel-on-rail systems.In all aspects,Maglev is superior to a conventional train.Table II represents the comparison of characteristics of the mass transportation systems provided by the Ministry of Trans-portation in Japan.It is appreciable from the tables that the ten-dency of global transportation is toward the Maglev train.Ac-cordingly,it is necessary to be concerned and understand all0018-9464/$20.00©2006IEEETABLE IIC OMPARISON OF C HARACTERISTICS OF THE M ASS T RANSPORTATION SYSTEMSparison of support,guidance,and propulsion.(a)Wheel-on-rail system.(b)Maglev system.technologies including magnetic levitation,guidance,propul-sion,power supply,and so on.II.T ECHNOLOGY A SPECTSState-of-the-art Maglev train technologies are investigated.Fig.1illustrates the difference between the conventional train and the Maglev train.While the conventional train uses a rotary motor for propulsion and depends on the rail for guidance and support,the Maglev train gets propulsion force from a linear motor and utilizes electromagnets for guidance and support.A.LevitationTypically,there are three types of levitation technologies:1)electromagnetic suspension;2)electrodynamic suspension;and 3)hybrid electromagnetic suspension.1)Electromagnetic Suspension (EMS):The levitation is ac-complished based on the magnetic attraction force between a guideway and electromagnets as shown in Fig.2.This method-ology is inherently unstable due to the characteristic of the mag-netic circuit [28].Therefore,precise air-gap control is indis-pensable in order to maintain the uniform air gap.Because EMS is usually used in small air gapslike 10mm,as the speed be-comes higher,maintaining control becomes dif ﬁcult.However,EMS is easier than EDS technically (which will be mentioned in Section II)and it is able to levitate by itself in zero or low speeds (it is impossible with EDS type).In EMS,there are two types of levitation technologies:1)the levitation and guidance integrated type such as Korean UTM and Japanese HSST and 2)the levitation and guidance sepa-rated type such as German Transrapid.The latter is favorable for high-speed operation because levitation and guidancedoFig.2.Electromagnetic suspension.(a)Levitation and guidance integrated.(b)Levitation and guidanceseparated.Fig.3.Electrodynamic suspension.(a)Using permanent magnets.(b)Using superconducting magnets.not interfere with each other but the number of controllers in-creases.The former is favorable for low-cost and low-speed op-eration because the number of electromagnets and controllers is reduced and the guiding force is generated automatically by the difference of reluctance.The rating of electric power supply of the integrated type is smaller than that of the separated type,but as speed increases,the interference between levitation and guidance increases and it is dif ﬁcult to control levitation and guidance simultaneously in the integrated type [29].In general,EMS technology employs the use of electromag-nets but nowadays,there are several reports concerning EMS technology using superconductivity,which is usually used for EDS technology [30]–[33].Development of the high-tempera-ture superconductor creates an economical and strong magnetic ﬁeld as compared with the conventional electromagnets even though it has some problems such as with the cooling system.2)Electrodynamic Suspension (EDS):While EMS uses attraction force,EDS uses repulsive force for the levitation [34]–[46].When the magnets attached on board move for-ward on the inducing coils or conducting sheets located on the guideway,the induced currents ﬂow through the coils or sheets and generate the magnetic ﬁeld as shown in Fig.3.The repulsive force between this magnetic ﬁeld and the magnets levitates the vehicle.EDS is so stable magnetically that it is unnecessary to control the air gap,which is around 100mm,and so is very reliable for the variation of the load.Therefore,EDS is highly suitable for high-speed operation and freight.However,this system needs suf ﬁcient speed to acquire enough induced currents for levitation and so,a wheel like a rubber tire is used below a certain speed (around 100km/h).By the magnets,this EDS may be divided into two types such as the permanent magnet (PM)type and the supercon-ducting magnet (SCM)type.For the PM type,the structure is very simple because there is no need for electric power supply.The PM type is,however,used for small systems only because ofLEE et al.:REVIEW OF MAGLEV TRAIN TECHNOLOGIES1919Fig.4.Hybrid electromagneticsuspension.Fig.5.Concept of the linear motor from the rotary motor.the absence of high-powered PMs.Nowadays,a novel PM such as the Halbach Array,is introduced and considered for use in the Maglev train (Inductrack,USA).For the SCM type,the struc-ture is complex,in addition,quenching and evaporation of liquid helium,which are caused from the generated heat of the in-duced currents,may cause problems during operation [49]–[60].Hence,helium refrigerator is indispensable for making the SCM operate.Nevertheless,the SCM type holds the world record of 581km/h in 2003in Japan.3)Hybrid Electromagnetic Suspension (HEMS):In order to reduce the electric power consumption in EMS,permanent mag-nets are partly used with electromagnets as illustrated in Fig.4[61]–[67].In a certain steady-state air gap,the magnetic ﬁeld from the PM is able to support the vehicle by itself and the elec-tric power for the electromagnets that control the air gap can be almost zero.However,HEMS requires a much bigger vari-ation of the current ’s amplitude as compared with EMS from the electromagnets ’point of view because the PM has the same permeability as the air [68].B.PropulsionThe Maglev train receives its propulsion force from a linear motor,which is different from a conventional rotary motor;it does not use the mechanical coupling for the rectilinear move-ment.Therefore,its structure is simple and robust as compared with the rotary motor [69]–[71].Fig.5shows the concept of the linear motor derived from the rotary motor.It is a conven-tional rotary motor whose stator,rotor and windings have been cut open,ﬂattened,and placed on the guideway.Even though the operating principle is exactly the same as the rotary motor,the linear motor has a ﬁnite length of a primary or secondary part and it causes “end effect.”Moreover,the large air gap lowers the ef ﬁciency.However,the linear motor is superior to the rotary motor in the case of rectilinear motion,because of the less signi ﬁcant amount of vibration and noise that are generated directlyfromFig.6.Linear induction motor (LPtype).Fig.7.Linear synchronous motor (LP type).the mechanical contact of components such as the screw,chain,and gearbox.1)Linear Induction Motor (LIM):The operating principle of the LIM is identical to the induction motor.Space-time variant magnetic ﬁelds are generated by the primary part across the air gap and induce the electromotive force (EMF)in the secondary part,a conducting sheet.This EMF generates the eddy currents,which interact with the air-gap ﬂux and so produce the thrust force known as Lorenz ’s force.There are two types as follows.1)Short primary type (SP):stator coils are on board and con-ducting sheets are on the guideway.2)Long primary type (LP):stator coils are on the guideway and conducting sheets are on board as shown in Fig.6.For the LP type,construction cost is much higher than SP type but it does not need any current collector for operation.In high speeds,the LP type is usually used because transfer of energy using a current collector is dif ﬁcult.In the case of the SP type,it is very easy to lay aluminum sheets on the guideway and thereby reduce construction costs.However,the SP type has low energy ef ﬁciency because of the drag force and leakage inductance caused from the end effect.Secondly,the SP type cannot exceed around 300km/h on ac-count of the current collector.Therefore,the SP type LIM is generally applied for the low –medium speed Maglev trains such as the Japanese HSST or Korean UTM.2)Linear Synchronous Motor (LSM):Unlike the LIM,the LSM has a magnetic source within itself as shown in Fig.7.In-teraction between the magnetic ﬁeld and armature currents pro-duces the thrust force.The speed is controlled by the controller ’s frequency.According to the ﬁeld location,there are two types equivalent to the LIM (LP and SP type).Furthermore,there are another two types according to the magnetic ﬁeld.One of them utilizes the electromagnets with iron-core (German Transrapid)and the other uses the super-conducting magnets with air-core (Japanese MLX).High-speed Maglev trains prefer the LSM because it has a higher ef ﬁciency and power factor than the LIM.The economical ef ﬁciency of the electric power consumption is very important for high-speed operation.1920IEEE TRANSACTIONS ON MAGNETICS,VOL.42,NO.7,JULY2006Fig.8.Propulsion-guidance coils used in JapaneseMLU-002.Fig.9.Levitation-guidance coils used in Japanese MLX.Neither the LSM nor the LIM requires sensor techniques for their operation,and they are much alike in reliability and con-trollability but,as mentioned above,either one can be chosen based on speed,construction costs,and so on.C.GuidanceThe Maglev train is a noncontact system that requires a guiding force for the prevention of lateral displacement.As in the case of levitation,the guidance is accomplished electrome-chanically by magnetic repulsive force or magnetic attraction force [72]–[75].1)Using Magnetic Repulsive Force:As shown in Fig.8,by setting the propulsion coils on the left and right sides of the guideway and connecting the coils,the induced electromotive force (EMF)cancels out each other when the train runs in the center of the guideway.However,once a train runs nearer to one sidewall,currents ﬂow through the coils by the EMF induced by the distance difference.This produces the guiding force.In the MLX,by connecting the corresponding levitation coils of both sidewalls as shown in Fig.9,these coils work as a guide system.When a train displaces laterally,circulating currents be-tween these two coils are induced and this produces the guiding force.In the case of the Transrapid,lateral guidance electromag-nets are attached in the side of the vehicle and reaction rails are on both sides of the guideway.Interaction between them keeps the vehicle centered laterally as shown in Fig.12.2)Using Magnetic Attraction Force:As indicated in Fig.14,magnetic attraction force is generated in the way to reduce the reluctance and increase the inductance when the vehicle dis-places laterally.Because energy tends to ﬂow toward small re-luctance,this guides the vehicle centered laterally.Since guid-ance is integrated with levitation,the interference between them makes it dif ﬁcult to run at high speeds.Therefore,guidance using attraction force is used for low –medium speed operation such as the HSST orUTM.Fig.10.LSM design of Transrapid.(Linear generator is inserted in the levita-tion electromagnets).D.Transfer of Energy to VehicleEven though all Maglev trains have batteries on their vehi-cles,electric power supply from the ground side is necessary for levitation,propulsion,on-board electrical equipment,bat-tery recharging,etc.The transfer of energy all along the track involves the use of a linear generator or a mechanical contact based on the operation speed.1)Low–Medium Speed Operation:At low speeds up to 100km/h,the Maglev train,generally,uses a mechanical contact such as a pantograph.As has been pointed out,this is the reason why the SP type-LIM Maglev train is used for low –medium speed.2)High-Speed Operation:At high speeds,the Maglev train can no longer obtain power from the ground side by using a me-chanical contact.Therefore,high-speed Maglev trains use their own way to deliver the power to the vehicle from the ground [76],[77].The German Transrapid train employs the use of a linear generator that is integrated into the levitation electro-magnets as demonstrated in Fig.10.The linear generator de-rives power from the traveling electromagnetic ﬁeld when the vehicle is in motion.The frequency of the generator windings is six times greater than the motor synchronous frequency.The linear generator is mechanically contact-free,as aspect that is very positive for high-speed operation.However,ﬂuctuation of the induced voltage due to the unevenness of the airgap,and small magnitude of the induced voltage because of the minia-turized inducing coils can be a problem.For MLX,beside a gas turbine generator,two linear gener-ators are considered.The ﬁrst one utilizes exclusive supercon-ducting coils (500kA)and generator coils at the upper and lower sides as shown in Fig.11(a).The second one utilizes generator coils between superconducting coils and levitation-propulsion coils as shown in Fig.11(b).Because the ﬁrst one concentrates in the nose and tail of the vehicle,it is called the concentra-tion-type.The second one is known as the distribution-type be-cause it is distributed along the vehicle[101].With speed,these coils generate a variable ﬂux in the upper part of the levitation and guidance ﬁxed coils.Consequently,the lower part (generator coils)sees a variable ﬂux,which crosses the air gap.The variable ﬂux is coupled with on board generatorLEE et al.:REVIEW OF MAGLEV TRAIN TECHNOLOGIES1921Fig.11.Two types of the linear generators used in MLX.(a)Concentration-type.(b)Distribution-type.coils.In other words,a dcﬂux created by the on-board super-conducting coils is transformed in an acﬂux,on-board,via a linear transformer[101].III.W ORLDWIDE M AGLEV T RAIN P ROJECTSSince the Maglev train has been studied and developed from the1960s,both German and Japanese Maglev trains have reached industrial levels and test tracks are experienced.In the 1990s,the USA Inductrack,the Swiss Swissmetro,and Korea’s UTM have been intensively studied and some component pro-totypes have been built.The Transrapid in Shanghai,China and the HSST(High Speed Surface Transport)in Nagoya,Japan, have been in public service since December2003and March 2005,respectively.Some projects(Pittsburgh or Baltimore in USA,Seoul in Korea,London in England,and so on)are awaiting approval,and the Munich project in Germany was approved in September2003with public service possible from 2009[78]–[114].Tables III and IV represent the types and characteristics of the Maglev trains“in operation”and“in ready to use”states,respec-tively.The EDS levitation-type Maglev trains such as the MLU, MLX,and Inductrack,especially,need lateral and vertical wheel bogies to guide the vehicle at low speeds(below100km/h). There is one further thing that we cannot ignore.The MLX has higher maximum speed than the Transrapid.For the Transrapid, the maximum synchronous frequency is300Hz,which corre-sponds to limit of the power inverter.Such a limited frequency corresponds to a synchronous speed of around500–550km/h.TABLE IIIC LASSIFICATION OF THE M AGLEV T RAIN IN OPERATIONTABLE IVC LASSIFICATION OF THE M AGLEV T RAIN(R EADY TO U SE)Fig.12.Transrapid[107].However,for the MLX,superconducting technology permits a higher pole pitch(1350mm)than the Transrapid(258mm)and1922IEEE TRANSACTIONS ON MAGNETICS,VOL.42,NO.7,JULY2006Fig.13.Guideway of MLX[104].Fig.14.HSST.a corresponding lower synchronous frequency,72Hz can make 700km/h,which is the speed goal of the Japanese train [101].It is also notable that the low –medium speed Maglev train em-ploys SP-LIM as its propulsion type.Figs.12–14illustrate the Transrapid,infrastructure of the MLX,and the HSST system,respectively.Fig.15represents the diagram of the development of the global Maglev trains in chronological order.IV .C ONCLUSIONThe Maglev train is considered for both urban transportation and intercity transportation systems.In the low –medium speed Maglev train,the operating routine is shorter than the high-speed train.Therefore,EMS technology and LIM is preferred from the construction cost viewpoint.However,in high-speed operation,EDS technology and LSM is preferred for controlla-bility and reliability.In addition,as along with the development of the high temperature superconductor and new type of perma-nent magnets,stronger magnetic energy that is more cost effec-tive will be used for the Maglev train.Authors are sure that this technology can be utilized for not only train application but also aircraft launching systems and spacecraft launchingsystems.Fig.15.Development diagram of the global Maglev train.The need for a new and better transportation system has en-couraged many countries to be interested in and attempt to de-velop the Maglev train.However,even though the Maglev train has been studied and developed for approximately half a cen-tury,only a few countries have the knowledge and expertise to do so.This review paper tried to describe the present complete system in detail and summarize foundational core technologies of the Maglev train from an electrical engineering point of view.It is certain that this review paper will be helpful for persons who are interested in this matter to assimilate the Maglev train tech-nologies including magnetic levitation,propulsion,guidance,and power supply system.It only remains to be said that besides core technologies,there is still the need to obtain a better understanding of how various factors may in ﬂuence the system.For example,the dynamic be-havior of the vehicle with the in ﬂuence of the guideway may cause the mechanical dynamic resonance phenomena;air vibra-tion rattles the windows of buildings near tunnel portals when a Maglev train enters or leaves a tunnel at high speed;the pas-senger safety issue is not considered fully;vehicle vibration generated from the rough guideway construction also remains.And furthermore,cost-effectiveness is still undecided.A CKNOWLEDGMENTThe authors would like to thank Dr.Y.Lee and Dr.S.Lee,Korea Railroad Research Institute (KRRI),for their support.R EFERENCES[1]S.Yamamura,“Magnetic levitation technology of tracked vehiclespresent status and prospects,”IEEE Trans.Magn.,vol.MAG-12,no.6,pp.874–878,Nov.1976.[2]P.Sinha,“Design of a magnetically levitated vehicle,”IEEE Trans.Magn.,vol.MAG-20,no.5,pp.1672–1674,Sep.1984.[3]D.Rogg,“General survey of the possible applications and developmenttendencies of magnetic levitation technology,”IEEE Trans.Magn.,vol.MAG-20,no.5,pp.1696–1701,Sep.1984.[4]A.R.Eastham and W.F.Hayes,“Maglev systems development status,”IEEE Aerosp.Electron.Syst.Mag.,vol.3,no.1,pp.21–30,Jan.1988.[5]E.Abel,J.Mahtani,and R.Rhodes,“Linear machine power require-ments and system comparisons,”IEEE Trans.Magn.,vol.14,no.5,pp.918–920,Sep.1978.LEE et al.:REVIEW OF MAGLEV TRAIN TECHNOLOGIES1923[6]J.Fujie,“An advanced arrangement of the combined propulsion,levi-tation and guidance system of superconducting Maglev,”IEEE Trans.Magn.,vol.35,no.5,pp.4049–4051,Sep.1999.[7]P.Burke,R.Turton,and G Slemon,“The calculation of eddy losses inguideway conductors and structural members of high-speed vehicles,”IEEE Trans.Magn.,vol.MAG-10,no.3,pp.462–465,Sep.1974. [8]B.-T.Ooi,“Electromechanical dynamics in superconducting levitationsystems,”IEEE Trans.Magn.,vol.MAG-11,no.5,pp.1495–1497, Sep.1975.[9]G.Bohn,“Calculation of frequency responses of electro-magneticlevitation magnets,”IEEE Trans.Magn.,vol.MAG-13,no.5,pp.1412–1414,Sep.1977.[10]S.Aoki,“3-dimensional magneticﬁeld calculation of the levitationmagnet for HSST by theﬁnite element method,”IEEE Trans.Magn., vol.MAG-16,no.5,pp.725–727,Sep.1980.[11]R.Knowles,“Dynamic circuit and Fourier series methods for momentcalculation in electrodynamic repulsive magnetic levitation systems,”IEEE Trans.Magn.,vol.MAG-18,no.4,pp.953–960,Jul.1982. [12]J.Williams,R.Paul,and J Simkin,“Three-dimensionalﬁnite-elementmodelling of a superconducting suspension system,”IEEE Trans.Magn.,vol.MAG-19,no.6,pp.2615–2618,Nov.1983.[13]K.Hayafune and E.Masada,“Dynamics of the PM type linear syn-chronous motor for magnetically levitated carrier vehicle,”IEEE Trans.Magn.,vol.MAG-23,no.5,pp.2578–2580,Sep.1987.[14]T.Takahashi and K.Kurita,“Computation of eddy currents induced ina conducting sheet under moving magnets,”IEEE Trans.Magn.,vol.24,no.1,pp.197–200,Jan.1988.[15]M.Tsuchimoto,K.Miya,A.Yamashita,and M.Hashimoto,“An anal-ysis of eddy current and Lorentz force of thin plates under moving mag-nets,”IEEE Trans.Magn.,vol.28,no.2,pp.1434–1437,Mar.1992.[16]T.Saitoh,N.Maki,T.Kobayashi,M.Shibata,and T.Takizawa,“Elec-tromagnetic force and eddy current loss in dynamic behavior of a su-perconducting magnetically levitated vehicle,”IEEE Trans.Appl.Su-percond.,vol.3,no.1,pp.417–420,Mar.1993.[17]J.L.He,D.M.Rote,and H.T.Coffey,“Applications of the dynamiccircuit theory to Maglev suspension systems,”IEEE Trans.Magn.,vol.29,no.6,pp.4153–4164,Nov.1993.[18]H.Fukumoto,Y.Kameoka,K.Yoshioka,T.Takizawa,and T.Kobayashi,“Application of3D eddy current analysis on magnetically levitated vehicles,”IEEE Trans.Magn.,vol.29,no.2,pp.1878–1881, Mar.1993.[19]M.Andriollo,G.Martinelli,A.Morini,and A.Tortella,“FEM calcu-lation of the LSM propulsion force in EMS-MAGLEV trains,”IEEE Trans.Magn.,vol.32,no.5,pp.5064–5066,Sep.1996.[20]D.Albertz,S.Dappen,and G.Henneberger,“Calculation of the in-duced currents and forces for a hybrid magnetic levitation system,”IEEE Trans.Magn.,vol.33,no.2,pp.1263–1266,Mar.1997. [21]K.Yoshida,J.Lee,and Y.J.Kim,“3-D FEMﬁeld analysis in con-trolled-PM LSM for Maglev vehicle,”IEEE Trans.Magn.,vol.33,no.2,pp.2207–2210,Mar.1997.[22]K.Davey,“Analysis of an electrodynamic Maglev system,”IEEETrans.Magn.,vol.35,no.5,pp.4259–4267,Sep.1999.[23]I.Takahashi and Y.Ide,“Decoupling control of thrust and attractiveforce of a LIM using a space vector control inverter,”IEEE Trans.Ind.Appl.,vol.29,no.1,pp.161–167,Jan.–Feb.1993.[24]D.Cho,Y.Kato,and D.Spilman,“Sliding mode and classical con-trollers in magnetic levitation systems,”IEEE Control Syst.Mag.,vol.13,no.1,pp.42–48,Feb.1993.[25]P.K.Sinha,F.B.Zhou,and R.S.Kutiyal,“Fault detection in elec-tromagnetic suspension systems with state estimation methods,”IEEE Trans.Magn.,vol.29,no.6,pp.2950–2952,Nov.1993.[26]J.F.Eastham,M.J.Balchin,P.C.Coles,and D.Rodger,“Comparisonof short primary linear machines for high speed maglev vehicles,”IEEE Trans.Magn.,vol.23,no.5,pp.2338–2343,Sep.1987.[27]J.F.Hoburg,“Modeling maglev passenger compartment static mag-neticﬁelds from linear Halbach permanent-magnet arrays,”IEEE Trans.Magn.,vol.40,no.1,pp.59–64,Jan.2004.[28]E.Riches,“Will Maglev lift off?,”IEE Review,pp.427–430,Dec.1988.[29]S.Kusagawa,J.Baba,K.Shutoh,and E.Masada,“Multipurpose de-sign optimization of EMS-type magnetically levitated vehicle based on genetic algorithm,”IEEE Trans.Appl.Supercond.,vol.14,no.2,pp.1922–1925,Jun.2004.[30]J.R.Hull,“Attractive levitation for high-speed ground transport withlarge guideway clearance and alternating-gradient stabilization,”IEEE Trans.Magn.,vol.25,no.5,pp.3272–3274,Sep.1989.[31]W.Jiasu,W.Suyu,R.Zhongyou,D.Xiaogang,L.Guobin,L.Jisan,Z.Cuifang,H.Haiyu,D.Changyan,and Z.Dequi,“A scheme of Ma-glev vehicle using high Tc bulk superconducters,”IEEE Trans.Appl.Supercond.,vol.9,no.2,pp.904–907,Jun.1999.[32]M.Tsuchiya and H.Ohsaki,“Characteristics of electromagnetic forceof EMS-type maglev vehicle using bulk superconductors,”IEEE Trans.Magn.,vol.36,no.5,pp.3683–3685,Sep.2000.[33]G.Zhang,Y.Fang,F.Song,G.Zhu,and Z.Wang,“Optimal designand FEM analysis of the superconducting magnets of EMS-MAGLEV models using Bi-2223tapes,”IEEE Trans.Appl.Supercond.,vol.14, no.2,pp.1850–1853,Jun.2004.[34]J.Hogan and H.Fink,“Comparison and optimization of lift and dragforces on vehicles levitated by eddy current repulsion for various null and normalﬂux magnets with one or two tracks,”IEEE Trans.Magn., vol.11,no.2,pp.604–607,Mar.1975.[35]T.Iwahana,“Study of superconducting magnetic suspension and guid-ance characteristics on loop tracks,”IEEE Trans.Magn.,vol.11,no.6, pp.1704–1711,Nov.1975.[36]O.Tsukamoto,Y.Iwata,and S.Yamamura,“Analysis of ride qualityof repulsive type magnetically levitated vehicles,”IEEE Trans.Magn., vol.17,no.2,pp.1221–1233,Mar.1981.[37]R.J.Hill,“Teaching electrodynamic levitation theory,”IEEE Trans.Educ.,vol.33,no.4,pp.346–354,Nov.1990.[38]T.Sakamoto,A.R.Eastham,and G.E.Dawson,“Induced currents andforces for the split-guideway electrodynamic levitation system,”IEEE Trans.Magn.,vol.27,no.6,pp.5004–5006,Nov.1991.[39]J.L.He,Z.Wang,D.M.Rote,and S.Winkelman,“Investigation ofthe stability of AC repulsive-force levitation systems for low-speed ma-glev,”IEEE Trans.Magn.,vol.28,no.5,pp.3315–3317,Sep.1992.[40]E.E.Burkhardt,J.Schwartz,and S.Nakamae,“Analysis of super-conducting magnet(SCM)-ground coil interactions for EDS Maglev coil conﬁgurations,”IEEE Trans.Appl.Supercond.,vol.3,no.1,pp.430–433,Mar.1993.[41]E.Leung,M.Dew,G.Samavedam,and B.Gamble,“A study of twodistinct coil designs for a Maglev EDS application,”IEEE Trans.Magn.,vol.30,no.4,pp.2379–2382,Jul.1994.[42]I.-Y.A.Wang and I.Busch-Vishniac,“A new repulsive magnetic lev-itation approach using permanent magnets and air-core electromag-nets,”IEEE Trans.Magn.,vol.30,no.4,pp.1422–1432,Jul.1994.[43]M.Andriollo,G.Martinelli,A.Morini,and A.Scuttari,“Optimizationof the winding conﬁguration in EDS-MAGLEV trains,”IEEE Trans.Magn.,vol.32,no.4,pp.2393–2398,Jul.1996.[44]D.-H.Kim,J.-K.Lee,S.-Y.Hahn,and G.Cha,“New levitation schemewith AC superconducting magnet for EDS MAGLEV system,”IEEE Trans.Magn.,vol.32,no.5,pp.5151–5153,Sep.1996.[45]K.R.Davey,“Electrodynamic Maglev coil design and analysis,”IEEETrans.Magn.,vol.33,no.5,pp.4227–4229,Sep.1997.[46]D.M.Rote and Y.Cai,“Review of dynamic stability of repulsive-forcemaglev suspension systems,”IEEE Trans.Magn.,vol.38,no.2,pp.1383–1390,Mar.2002.[47]J.de Boeij,M.Steinbuch,and H.M.Gutiérrez,“Mathematical modelof the5-DOF sled dynamics of an electrodynamic Maglev system witha passive sled,”IEEE Trans.Magn.,vol.41,no.1,pp.460–465,Jan.2005.[48]——,“Modeling the electromechanical interactions in a null-ﬂux elec-trodynamic Maglev system,”IEEE Trans.Magn.,vol.41,no.1,pp.466–470,Jan.2005.[49]H.Tsuchishima and T.Herai,“Superconducting magnet and on-boardrefrigeration system on Japanese MAGLEV vehicle,”IEEE Trans.Magn.,vol.27,no.2,pp.2272–2275,Mar.1991.[50]M.E.Vermilyea and C.Minas,“A cryogen-free superconductingmagnet design for Maglev vehicle applications,”IEEE Trans.Appl.Supercond.,vol.3,no.1,pp.444–447,Mar.1993.[51]E.A.Scholle and J.Schwartz,“Power dissipation due to vibration-in-duced disturbances in maglev superconducting magnets,”IEEE Trans.Appl.Supercond.,vol.4,no.4,pp.205–210,Dec.1994.[52]M.Andriollo,G.Martinelli,A.Morini,and A.Scuttari,“Minimiza-tion of the induced current effects in the shields of SC coils in EDS-MAGLEV trains,”IEEE Trans.Appl.Supercond.,vol.5,no.2,pp.604–609,Jun.1995.[53]S.Yokoyama,K.Shimohata,T.Inaguchi,T.Takeuchi,T.Kim,S.Nakamura,S.Miyashita,and F.Uchikawa,“A conceptual design of a superconducting magnet for MAGLEV using a Bi-based high-Tc tape,”IEEE Trans.Appl.Supercond.,vol.5,no.2,pp.610–613,Jun.1995.[54]A.Ametani,R.Kato,H.Nishinaga,and M.Okai,“A study of transientinduced voltages on a Maglev train coil system,”IEEE Trans.Power Del.,vol.10,no.3,pp.1657–1662,Jul.1995.。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。