P1U1networkn.网络,电路resistorn.电阻器inductorn.电感器capacitorn.电容器passivenetwork无源网络activenetwork有源网络characteristicadj.特性(的);n.特性曲线Ohmn.欧姆Faradayn.法拉第electriccharge电荷integraln.积分incrementn.增量armaturen.电枢,衔铁,加固aforementionedadj.上述的,前面提到的representv.代表,表示,阐明amplifyv.放大symbolicadj.符号的,记号的meshn.网孔Kirchhoff’sfirstlaw基尔霍夫第一定律loopcurrent回路电流voltagedrop电压降inseries串联differentialadj.微分的;n.微分variablen.变量outlinen.轮廓;v.提出……的要点eliminatev.消除,对消Pulsatev.脉动,跳动,振动apparatusn.一套仪器,装置ratedadj.额定的,设计的,适用的coiln.绕组,线圈;v.盘绕distributionn.分配,分布,配电generatorn.发生器,发电机emf(electromotiveforce)电动势interconnectv.互相连接wyen.Y形联结,星形联结,三通deltan.希腊字母△,三角形(物)geometryn.几何学,几何形状windingadj.缠绕的;n.线圈,绕组polarityn.极性neutraladj.中性的;n.中性线subscriptn.下标,角注,索引succeedv.继……之后,接替intersectionn.相交,逻辑乘法phasesequence相序reversev.,n.反转;adj.变换极性的P1U2amplifiern.放大器integratedcircuit集成电路buildingblocks积木potentialn.(电)势cascaden.,v.串联;adj.串联的ontheorderof属于同类的,约为tradeoff换取cumbersomeadj.麻烦的intrinsicadj.内在的circuitryn.电路transistorn.晶体管semiconductorn.半导体dopev.掺入predominancen.优势crystaln.晶体germaniumn.锗silicon硅bipolar双向的lead引线substitution代替detector探测器bias偏压polarity极性encompass包含moreorless或多或少cylindrical圆柱形的can密封外壳triangular三角的elongate延长,拉长P1U3flip-flopn.触发器relevancen.关联terminologyn.术语aptnessn.恰当pilotn.飞行员aloftadv.高高地cockpitn.坐舱deducev.演绎simultaneouslyadv.同时地Booleanalgebra布尔代数gaten.门,门电路prevalentadj.流行的inhibitv.抑制binaryadj.二进制的paralleln.类似decimaladj.十进制的radixn.权chainn.串remaindern.余数igitn.位数fractionaladj.小数的hexadecimaladj.十六进制的octaladj.八进制的alignmentn.组合p1U4convertern.转换器,换流器,变流器matrixn.模型,矩阵dioden.二极管,半导体二极管thyristorn.晶闸管triacn.三端双向晶闸管GTO门极可关断晶闸管BJT双极结型晶体管powerMOSFET电力MOS场效应晶体管SIT静态感应晶体管IGBT绝缘栅双极型晶体管MCTMOS控制晶闸管IGCT集成门极换向晶闸管rectificationn.整流feedbackn.反馈freewheelingn.单向传动snubbern.缓冲器,减震器intrinsicadj.固有的,体内的,本征forwardbiased正向偏置conductionn.导电,传导emittern.发射极reversebiased反向偏置leakagecurrent漏电流thresholdn.门限,阈限,极限breakdownn.击穿,雪崩recoveryn.恢复schottkydiode肖基特二极管workhorsen.重载,重负荷thyratronn.闸流管breakovern.导通latchingcurrent闭锁电流holdingcurrent保持电流phasecontrolled相控的asymmetricadj.不对称的forcecommutated强制换向SMPS开关电源BLDM无刷直流电动机steppermotor步进电动机hybridn.混合saturationn.饱和rectifiern.整流器choppern.斩波器invertern.逆变器cycloconvertern.周波变换器electrochemicaladj.电化学的VAR静态无功功率harmonicsn.谐波laggingn.滞后,迟滞powerfactor功率因数configurationn.轮廓,格局voltage-fedinverter电压源型逆变器current-fedinverter电流源型逆变器stiffvoltagesource恒压源stiffcurrentsource恒流源Theveninimpedance戴维南电路等效阻抗filtern.滤波器isolationtransformer隔离变压器buckchopper降压式变压器boostchopper升压式变压器quadrantn.象限dutyratio占空比,功率比P2U1regulatev.调整aboundv.大量存在powerboost功率助推装置aerodynamicadj.空气动力学的dampv.阻尼,减幅,衰减yawn.偏航altituden.海拔attituden.姿态intuitionn.直觉trail-and-errorn.试凑法dynamicresponse动态响应disturbancen.扰动parametern.参数modificationn.修正,修改transferfunction传递函数domainn.域,领域adventn.出现statevariable状态变量matrixalgebra矩阵代数approachn.途径,方法;研究proponentn.提倡者detractorn.批评者tutorialadj.指导性的subsequentadj.后序的open-loopn.开环closed-loopn.闭环discreteadj.离散的differentialequation微分方程differenceequation差分方程intervaln.间隔sampled-datan.采样数据nonlinearadj.非线性的time-invariantadj.时不变的coefficientn.系数stationaryadj.静态的lumpedparameter集中参数distributedparameter分散参数spatialadj.空间的springn.弹簧leadn.导线resistancen.阻抗uniformadj.一致的elasticadj.有弹性的ordinarydifferentialequation常微分方程partialdifferentialequation偏微分方程deterministicadj.确定的stochasticadj.随机的predictableadj.可断定的probabilitytheory概率论P3U5embeddedsystem嵌入式系统antilockbrakingsystem防抱死系统equivalentadj.相等的,相当的attributen.品质,特征high-endn.高端preferencesn.参数选择solidoutput可靠输出automatictellermachine自动柜员机answeringmachine电话答录机thermostatn.自动调温器Inertialguidancesystems惯性导航系统aircraftn.飞行器missilen.导弹sprinklern.洒水车,洒水装置infraredadj.红外线的reinventv.彻底改造VideoCassetteRecorder(VCR)录像机hydraulicadj.水压的exponentialadj.指数的defrostv.除霜deodorizev.除……臭lapsen.(时间等)流逝drasticadj.激烈的determinismn.决定论deterministicadj.确定性的redundancyn.冗余tacticn.策略,战略domainn.范围,领域摘要:针对直升机飞行状态识别训练样本数据少,导致识别率不高的问题,提出一种基于SVM的直升机飞行状态识别方法。首先,利用限幅、去野点和均值滤波方法对飞行数据进行去噪;其次,利用最小二乘法对飞行数据进行直线拟合获取变化率,并利用线性相关性提取状态特征参数,减少数据冗余;然后,根据特征参数对飞行状态进行预分类,分为十个小类,并对每一小类进行SVM分类器设计,提高识别效率;最后,利用训练样本对每一个SVM分类器进行训练,并用训练好的SVM分类器完成直升机全起落飞行状态的识别。通过某型直升机实飞数据进行飞行状态识别实验,并将本文方法与RBF神经网络法进行对比,实验结果表明,在小样本情况下,本文方法识别率有明显的提高,为直升机故障诊断和寿命预测提供重要依据。Abstract:Tosolvetheproblemoflowrecognitionrateduetotheshortageoftrainingsamples,aflightconditionrecognitionmethodbasedonSVMwasproposed.Firstly,theflightdatawasdenoisedbyclipping,removingtheoutlierdataandaveragefiltering.Secondly,changerateofflightdatawasobtainedbyleastsquarelinefitting,anddataredundancywasdecreasedbylinearindependencyusedtoextractthecharacteristicsparameters.Thirdly,flightconditionwasdividedintotenclassesbycharacteristicsparametersandSVMclassifierwasdesignedforeachclasssoastoimprovetheidentificationefficiency.Finally,everySVMclassifierwastrainedbytrainingsamples,andallflightconditionofhelicopterwasidentifiedbytrainedSVMclassifier.TheactualflightexperimentsshowthatcomparewithRBFneuralnetworkmethod,thismethodcouldimprovetherecognitionrateintheconditionofsmallsample,providingareferenceforhelicopterinfaultdiagnosesandlifeprediction.