rm[].ThebasesoftheWaveletTransformarethewavelets,generatedfromabasicwaveletfunctionbydilationsandtranslations.Givenatime-varyingsignalf(t),awavelettransformationconsistsofcomputingcoefficientsthataretheinnerproductsofthesignalandafamilyofwavelets[].ByDWTweunderstandthecontinuouswaveletswiththediscretescaleandtranslationfactors[].TheDWTisdefinedasstatedinEq.(),wherecj,kiscalledthewaveletcoefficientundingnoise,whichisinherenttothecuttingprocesses,astheacousticandultrasonicvibrations.Thetechniquesmorefrequentlyreportedarerelatedtotorque,forceandfeedrate,sincecuttingforcesincreasewhenthetoolwearincreases[,].Spindlemotorcurrentandfeeddrivercurrentarecloselyrelatedtotheforcesinvolvedinmachiningsimilartotorquemeasurement[],sincebothshowtheamountofconsumedpowerinthecuttingprocess.Drivercurrentmonitoringisthebestapproachtoacquiresignalswithoutsensorsbecauseinthiswaythemachineisnotmodified,evenwhencurrentsensorsareused[].Inaddition,thecurrentfromservomotorsisavailableinalmostallmodernrotatingmachineryasturning,millinganddrilling,directlyfromtheservodriver.Practicallyallthecurrentanalysisbasedalgorithmsreporteduseasensortoobtainthecuttingsignals..TheoreticalbackgroundAccordingtoAltintas[],themaincomponentsofcuttingforces(dFt,dFf,dFr)canbeevaluatedinthex,y,andzdirections.ThetotalthrustanddrilltorquecanbeevaluatedbythesumofthecontributionofallMlipelements.Thetotalthrustforceisfoundbyaddingthecontributionsofthechiselandlipsthatcanbeseenasasinusoidalfunctionplusaconstant.Directmeasurementsofspindlecurrentsignalshowsevereinterferencefromdifferentsourcesrepresentedintheadditivemodel(Fig.).TheapproachofRomero-Troncoso[]isageneralproceduretoestimatethemaincomponentsofthecuttingforcewhichextractsthefiltercharacteristics.Byapplyingthisapproachtothedrillingproblemweusealow-passfilter(LPF)toensurethatthespectralcontentsofthecuttingsignalarepreserved,whilespuriousdataareminimized.Thedesignedfilterdoesnoteliminateallspuriouscomponents,butasubsequentDiscreteWaveletTransform(DWT)willenhancethecuttingforcesignalbyitsfilterbankproperty.Thewavelettransformbringsatime-frequencyrepresentationofasignalindecimatedformdependingontheapplicationdetaillevel;theresultwillbegivenastimedomainsamplesatthedecimatedfrequencyinacompressedform[].ThebasesoftheWaveletTransformarethewavelets,generatedfromabasicwaveletfunctionbydilationsandtranslations.Givenatime-varyingsignalf(t),awavelettransformationconsistsofcomputingcoefficientsthataretheinnerproductsofthesignalandafamilyofwavelets[].ByDWTweunderstandthecontinuouswaveletswiththediscretescaleandtranslationfactors[].TheDWTisdefinedasstatedinEq.(),wherecj,kiscalledthewaveletcoefficient正常刀具原始切削信号(b)正常刀具过滤切削信号(c)正常刀具原始数据段(d)正常刀具压缩数据段图B.(a)已损刀具原始切削信号(b)已损刀具过滤切削信号(c)已损刀具原始数据段(d)已损刀具压缩数据段图(a)-(c)说明了当钻床处于较好状态时,对每秒一节原始信号所有分析。在这种切削状况下,工件粗糙度为.微米。在图(a)中,由交直流叠加原始电流信号从主轴驱动电机中获取。这类信号充满了噪声,使刀具状况可靠分析变得困难。图(b)表明了无高频成分原始信号周期形式,这是因为用一低通滤波器把它们滤掉了。为了估计钻头疲劳程度,新噪声限制了信号可以被处理。根据实验,信号中直流级别被钻头进给率和工件硬度所影响。为了编写非对称算法,有必要在信号中分析两组连续脉冲。在图(c)中,两组脉冲由给出在每一脉冲波形变化决定。为了计算这种变化,为了确保更快处理,减少点数量(通过小波变化来压缩信号)。图(d)表明,应用四次小波变换来保存原始无干扰信号波形,但是用很少数据定义。此时,利用自相关函数方程()我们能轻易而快速地估算两组脉冲无规律情况。图(d)说明了一些样本对比,这些信号决定了非对称值部分;对于这个特别区段,它是.个单位。图(a)-(d)表明像图那样相似分析,但是在这种情况下,刀具侧面疲劳量为.毫米,粗糙度为.微米。图(a)描绘了原始噪声信号。靠在秒数据区应用一低通滤波器,一种新信号产生了。在这种情况下,无规律波形反应了钻头疲劳情况。在图(c)中,如果有足够关于钻头疲劳率信息,两组脉冲信号有不对称形式。图(d)说明了在脉冲波形上巨大差异,和疲劳和不对称值都很小好刀具例子不同;因此,.个单位不对称值是可以达到。图B.(a)正常刀具不对称区值(b)已损刀具不对称区值两组连续脉冲比较继续通过所有切削信号,在这种方式下,一种全面疲劳模式依照由不对称呈现信号变形产生。不对称信号中点数量将由钻切循环时间决定。每两个脉冲可计算出一个新值,所以当工具达到一定固定疲劳水平时,加工应在错误状况出现之前按时停止,以防损坏刀具和机床。图(a)和(b)分别为数据秒展示出了不对称值,数据来自图(a)和(a)。这种方法先进性是疲劳鉴定只是基于连续脉冲变形,对监测系统来说,这是一种非常快刀具分析。图B.(a)正常刀具非对称区(b)疲劳刀具非对称区(c)断裂刀具非对称区图B.钻孔时非对称区结论不对称值将会依照刀具情形而增加。疲劳率越高,切削信号不对称度越大。由于钻头尺寸,在刀具移位之前通常会有重要侧面疲劳。对允许小不对称值来说,工件质量由机床上实验决定。为了找到被认可最大值而应用一些钻头。图(a)-(c)说明三种钻头疲劳不同比例和它们每段数据对应不对称值。在图(a)中,刀具很锋利并且有很小连续不对称值,总是低于个单位;在最先个周期,这种过程被观测。在周期中,测试刀具显示.毫米,侧面疲劳见图(b)。对这种程度疲劳,不对称度增加个单位,即使被加工工件表现出很好质量;在接下来个周期中这个不对称度将会达到临界值。对机床来说最糟糕情况见图(c)中。.毫米侧面疲劳产生了高出平均值个单位不对称值,这和有缺陷工件有关。在周期末,这种情况继续恶化,成为了更糟随后孔。在这样环境下,孔粗糙度(.微米)和生产套尺度都曾被认为质量部不足。包括所有钻削周期不对称值图表说明见图。这会再现钻过程和它和波形不规律近似关系。既然是在第一个周期,不对称增长很明显。对第一个周期来说,对大值仅仅在两个单位左右,但是在周期之前,不对称值在连续地、缓慢地增长。到达钻削周期时,工件已加工表面质量和直径大小超出了必要限制。所以,更高不对称度是不可接受,刀具更换是被推荐。结束语主轴驱动电流也许跟钻削工艺动力学和切削刀具监测有关。小波变换很有用,因为它靠对原始切削力信号压缩和滤波来简化了信号分析处理。不对称算法有很高可靠性可以区分开已疲劳钻头和可用钻头,并且提供在不同钻削状况下钻头疲劳精确计算。所有算法已经作为处理单元在PC机上测试,但是为了得到控制机床在线监测系统,它用SOC(芯片系统)来实现;因此,系统是无传感器并且利用伺服驱动器电流。靠补偿公差参数,它能根据监测出疲劳类型而用在不同设备上。PC机将不再成为必需,因此,这种推荐方法将有效节约成本。在市场上,有很多商业设备用在侦测刀具疲劳问题,但是既然他们利用一两个传感器,那解决方案就贵了起来。我们建议是一套种无传感器系统,它能在无电脑系统下工作并且可以在线应用。undingnoise,whichisinherenttotheuttingprocesses,astheacousticandultrasonicvibrations.Thetechniquesmorefrequentlyreportedarerelatedtotorque,forceandfeedrate,sincecuttingforcesincreasewhenthetoolwearincreases[,].Spindlemotorcurrentandfeeddrivercurrentarecloselyrelatedtotheforcesinvolvedinmachiningsimilartotorquemeasurement[],sincebothshowtheamountofconsumedpowerinthecuttingprocess.Drivercurrentmoni中文字附录A外文翻译原文SensorlesstoolfailuremonitoringsystemfordrillingmachinesLuisAlfonsoFranco-Gascaa,GilbertoHerrera-Ruiza,RocıoPeniche-Veraa,RenedeJesusRomero-Troncosob,WbaldoLeal-Tafollac出处:InternationalJournalofMachineToolsandManufactureVolume,Issues–,March,Pages–AbstractItiswellknownthaton-linetoolconditionmonitoringhasgreatsignificanceinmodernmanufacturingprocesses.Inordertopreventpossibledamagestotheworkpieceorthemachinetool,reliabletechniquesarerequiredprovidinganon-lineresponsetoanunexpectedtoolfailure.Drillingisoneofthemostfundamentalmachiningoperationsandtwoofthemostcrucialissuesrelatedtoitaretoolwearandfracture.Duringthespindleprocess,themotordrivercurrentisrelatedtothedrillcondition:powerconsumptionishigherforaworndrillincomparisontoasharpdrillforthesameprocess.Thisdifferenceinpowerconsumptioncanbeself-correlatedtoobtaintheresultingwaveformvariancetoprovideameritfigurefortoolcondition.ThispaperdescribesadrivercurrentsignalanalysistoestimatethetoolconditionbyusingthediscreteWaveletTransforminordertoextracttheinformationfromtheoriginalcuttingforce,andthroughanautocorrelationalgorithmevaluatethetoolwearintheformofanasymmetryweightingfunction.Thecurrentismonitoredfromthemotordrivertogiveasensorlessapproach.Experimentalresultsarepresentedtoshowthealgorithmperformance,acompletesensorlesstoolfailuresystemwhichallowsthedetectionoftoolfailureasafunctionofspindlecurrentinrealtime.Keywords:Toolfailure;Wavelettransform;Toolmonitoring.IntroductionCuttingtoolsrepresentthehighestcostintheproductionprocessofthemanufacturingindustrybesidesrawmaterials.Drillingisoneofthemostcommonoperationsinmachining,thusthenumberofdrillingmachinesinuseperformaconsiderableamountofworkeveryyear[].Asinanyothercuttingprocesses,toolfractureandweararepresentinseveralforms;therefore,theycoulddamagetheprocessandincreasetheproductioncosts.Toreduceexpensesontheworkpieceandmachinery,on-linetoolconditionmonitoringismandatory[].Aroundofthereportedcasesreturntheinvestmentinmonitoringsystemsinonemonthorless[].Jantunen[]describesthestudiedandappliedtechniquesofindirectmonitoringindrilling.Vibrationandsoundbasedworkshavebeenreported.Suchmethodsareverysensitivetosurroundingnoise,whichisinherenttothecuttingprocesses,astheacousticandultrasonicvibrations.Thetechniquesmorefrequentlyreportedarerelatedtotorque,forceandfeedrate,sincecuttingforcesincreasewhenthetoolwearincreases[,].Spindlemotorcurrentandfeeddrivercurrentarecloselyrelatedtotheforcesinvolvedinmachiningsimilartotorquemeasurement[],sincebothshowtheamountofconsumedpowerinthecuttingprocess.Drivercurrentmonitoringisthebestapproachtoacquiresignalswithoutsensorsbecauseinthiswaythemachineisnotmodified,evenwhencurrentsensorsareused[].Inaddition,thecurrentfromservomotorsisavailableinalmostallmodernrotatingmachineryasturning,millinganddrilling,directlyfromtheservodriver.Practicallyallthecurrentanalysisbasedalgorithmsreporteduseasensortoobtainthecuttingsignals..TheoreticalbackgroundAccordingtoAltintas[],themaincomponentsofcuttingforces(dFt,dFf,dFr)canbeevaluatedinthex,y,andzdirections.ThetotalthrustanddrilltorquecanbeevaluatedbythesumofthecontributionofallMlipelements.Thetotalthrustforceisfoundbyaddingthecontributionsofthechiselandlipsthatcanbeseenasasinusoidalfunctionplusaconstant.Directmeasurementsofspindlecurrentsignalshowsevereinterferencefromdifferentsourcesrepresentedintheadditivemodel(Fig.).TheapproachofRomero-Troncoso[]isageneralproceduretoestimatethemaincomponentsofthecuttingforcewhichextractsthefiltercharacteristics.Byapplyingthisapproachtothedrillingproblemweusealow-passfilter(LPF)toensurethatthespectralcontentsofthecuttingsignalarepreserved,whilespuriousdataareminimized.Thedesignedfilterdoesnoteliminateallspuriouscomponents,butasubsequentDiscreteWaveletTra 中文4750字附录A外文翻译原文SensorlesstoolfailuremonitoringsystemfordrillingmachinesLuisAlfonsoFranco-Gascaa,GilbertoHerrera-Ruiza,Rocı´oPeniche-Veraa,Rene´deJesu´sRomero-Troncosob,WbaldoLeal-Tafollac出处:InternationalJournalofMachineToolsandManufactureVolume46,Issues3–4,March2006,Pages381–386AbstractItiswellknownthaton-linetoolconditionmonitoringha