1、12FTM09AGMA Technical PaperSystematic Approachfor the PsychoacousticAnalysis of DynamicGear Noise ExcitationBy C. Brecher, M. Brumm, andC. Carl, RWTH Aachen UniversitySystematic Approach for the Psychoacoustic Analysis ofDynamic Gear Noise ExcitationDr.ChristianBrecher,MarkusBrumm,andChristianCarl,R
2、WTHAachenUniversityThe statements and opinions contained herein are those of the author and should not be construed as anofficial action or opinion of the American Gear Manufacturers Association.AbstractThe sound quality of technical products is an increasingly important quality criterion and has a
3、significantinfluenceontheproductacceptance. Butsoundqualitydoesnotonlydependonthephysicalattributesofthesoundsignal. Itisdefinedtoalargeextentbyhumansoundandnoiseperception. Thisperceptionisbasedona physiological and psychological signal processing. These aspects depend on complex properties of thep
4、hysical signal like the spectral distribution and a relative comparison. However, today the sound design ofgearboxes is mainly based on the physical reduction of the noise level that is detected by absolute andobjectivized parameters. The noise oriented gear design is based on physical key parameter
5、s like thereductionoftransmissionerrorincompliancewithachievablemanufacturingtolerances. Nevertheless,thesedesign rules may lead to a minimal sound pressure level but cannot solely be applied for an optimal soundqualityineverycase. Undereconomicandtechnicalaspectsthereisnoexcitationfreegearset. Furt
6、hermore,modern tendencies such as lightweight design and masking noise reduction (engine downsizing andelectrification) lead more and more to scenarios where the sound of a gear set, which is only designated tohave low transmission error, can be perceived as annoying. This requires design guidelines
7、 which take alsothe human related aspects of gear noise into account. Nowadays the gear design does not yet considerhuman noise perception sufficiently.Thus, a research project at the WZL has been established that investigates the correlation between gearmesh excitation and the evaluation of gear no
8、ise. The objective of this project is to deduce a method for theconsideration of perception-based noise evaluation already in the stage of gear design. Therefore,psychoacoustics metrics are used to analyze the gear noise of different gear sets in the dimensions ofair-bornenoise,structuralvibrationan
9、dtheexcitationduetomeshing. Theaimof thispaper isto discussthecorrelation between the signal properties of the excitation and the radiated noise in order to investigate thepossibilities to transfer the perception related evaluation from sound pressure to the gear mesh excitation.The paper firstly sh
10、ows central psychoacoustic parameters that are most relevant for the properties of gearnoise. Furthermore, a new test fixture will be introduced that allows a dynamic measurement of gear meshexcitation directly adjacent to the meshing. Regarding these aspects two different gear sets are discussedcon
11、cerningthe calculatedtransmission error and theexperimentallydeterminedexcitation,surfacevibrationandnoiseradiation. Theseaspectsareaccordinglyexaminedwithrespecttohumannoiseperception,whichisdescribedbypsychoacoustics. Itisshownthatoperatingconditions,orderdistributionsaswellasthegeargeometry are t
12、he main influences on the signal evaluation. The influence of dynamic aspects and especiallythe influence of resonance effects on the noise characteristics are additionally considered.Copyright 2012American Gear Manufacturers Association1001 N. Fairfax Street, Suite 500Alexandria, Virginia 22314Octo
13、ber 2012ISBN: 978-1-61481-040-73 12FTM09Systematic Approach for the Psychoacoustic Analysis of Dynamic Gear NoiseExcitationDr. Christian Brecher, Markus Brumm, and Christian Carl, RWTH Aachen UniversityIntroduction and objectivesThe main objective of the gear design process is to meet the specificat
14、ions and requirements on fatigue andwear resistance considering the defined product life cycle. Nevertheless, beside this aspect the generatedand radiated noise of the gearbox becomes increasingly important in terms of acoustic quality and productacceptance by the customer. This leads to an increasi
15、ng demand for powertrains and gearboxes with a lowvibrational and acoustic output. In order to achieve and ensure a high quality rating acoustic design istherefore a central challenge for gearbox development.Today the acoustic development process of gearboxes is based on absolute and objectified ana
16、lyzingmethodsthatallowaphysicalminimizationofexcitationandvibration. Nevertheless,atotalreduction ofgearnoise cannot be achieved by reasonable effort taking into account practical requirements and manufacturingboundaryconditions. Incompliancewiththeseboundaryconditionsscenariosoccurwhenphysicalreduc
17、tionand optimization of the excitation cannot solely lead to an improved acoustic quality rating of the gearbox 12 3. The origin for the discrepancy between design and acoustic quality can be found in the signalprocessing of human noise perception that is not only determined by the physical sound pr
18、essure level butmore by its spectral distribution and psycho-physiological (psychoacoustic) aspects 4. Moreover, it can beobserved that improvementsof manufacturingquality mayeven leadto aworse acousticquality rating,sincethe reduction of randomly scattered excitation effects can lead to more emphas
19、ized tonal aspects of gearnoise 5 6 7. Therefore, an extended source-path-receiver approach is required for achieving optimalacoustic quality of a gearbox powertrain, Figure 1.This systematic approach needs to consider the entire gear noise origination process from the gear setsspecific properties t
20、o the noise radiation and its perception under dynamic operating conditions. This is anaugmented consideration of the conventional source-path-receiver model as described by 8. The quasi-static gear mesh excitation is the foundation for the radiated noise. It is a gear set specific dimension thatdep
21、endsonaspectslikeload,macroandmicrogeometry,manufacturingdeviationsandseveralothersmore.Figure 1. Augmented source-path-receiver model for gear noise excitation and perception4 12FTM09Today a gear set is designed with the objective to have a low transmission error as an indicator for low noiseexcita
22、tion,whichismainlydistributedindiscretetonalharmonicsofthegearmeshfrequency. Itcanbequalitycontrolled by e.g., topographic measurements or rolling tests. At operating conditions interactions betweenthegearsetexcitationandeigendynamicsofthepowertrainleadtoadynamicforceexcitationthatistransmit-ted by
23、structural vibrations in the shaft-bearing-housingsystem andradiated from free surfaces as airbornenoise. The powertrain noise is evaluated by human perception as a result of aphysiologic andpsychologicalsignalprocessing. Todayalreadyseveralcalculationparametersareavailableandalsopartiallystandardiz
24、edthat allow considering this signal processing by objectified metrics. These psychoacoustic metrics can beused for an acoustic target design of powertrains 9 10 11 12 and also as a detection tool for conditionmonitoring objectives 13.However, today a method is not yet available for an adequate cons
25、ideration of human noise perception interms of the acoustically oriented gear design. Thus, the aim of this research activity is to analyze theperception related correlation between gear mesh excitation, structural vibrations and radiated airbornenoise. This will allow defining a procedure to integr
26、ate the psychoacoustic signal processing into the geardesign process in order to develop gear transmissions that have an increased acoustic quality rating. Thispaper firstly discusses the transfer of psychoacoustic metrics on the signal properties along the augmentedsource-path-receiver model of two
27、 different recently tested gear sets. Thereby, an analysis of the physicalvibration signals is followed by a psychoacoustic discussion of these signals in an averaged way and atspecific operating conditions. The latter one shows how the excitation order distribution influences thepsychoacoustic rati
28、ng in interaction with dynamic aspects and the transfer path at operating conditions.Psychoacoustic evaluation parametersHuman perception of noise is neither objective nor absolute. A lot of the physiologic and psychologicalreasons are not yet entirely explored. But it can be determined that the per
29、ception is not only defined by thephysical level. Moreover, classifications of annoyance require a more sophisticated approach of signalanalysis. Psychoacousticsmetrics havebeen developmentin recentpast andalready partiallystandardizedthatallowevaluatinganairbornenoisesignalwithrespecttodifferentper
30、ceptionaspects. Thesemetricsrateasignalconsideringtheannoyanceandeveneuphonyregardingthoseperceptionaspects. Figure 2givesanoverview on central psychoacoustic metrics that are described in the following paragraphs.Figure 2. Overview on central psychoacoustic metrics5 12FTM09LoudnessLoudness is the p
31、sychoacoustic description of human intensity perception. It depends mainly on thefrequency content of the signal but also on the exposition duration 14. Effects of masking and critical band-widths take into account the interference between single tones and narrow bandwidth noise in the frequencydoma
32、in. The maximum of perception is for single sinusoidal tones at about 4 kHz. The curves of equalloudnessareshowninFigure 2. Aboveandbeneaththisfrequencyregiontheintensityperceptiondecreasessignificantly. In effect the sound pressure level needs to be higher there to cause the same intensity orloudne
33、ss perception as for 4 kHz.There are two different scales that describe loudness perception: phon and sone. They can be transferredinto each other in a non-linear way. The sone-scale is the actual psychoacoustic one,because doublingthesone-value means a doubling of intensity perception. The phon-sca
34、le is historically used for a technical rat-ing. Thephonandsoneratingscaleshaveincommonafixpointatasinusoidaltonewithafrequencyof1 kHzand a sound pressure level (SPL) of 40 dB. This sound is rated per definition with 1 phon and 1 sone.TheevaluationmethodforloudnessisstandardizedinDIN45631andDIN45631
35、/A1:201015. Thesupplement-ation A1 is for the loudness calculation of time-variant sounds. The calculation of loudness according toDIN 45631 is basedon amethod thatdetermines theloudness ofa complexnoise asan equivalentloudnessrating of a single tone. The calculation procedure can be divided into th
36、ree steps. In the first step the fre-quencyspectrumofthenoiseisdissectedintocriticalfrequencygroupsofhearing andweighted accordingtothe frequency dependency of loudness. In the second step the effect of masking thresholds are consideredanda frequencygroup dependentcurve ofspecific loudnessN(z) isdet
37、ermined. In thelast stepthis curveisintegratedacrossallthefrequencygroups. TheresultingintegralcorrespondstotheloudnessNoftheentire,complex noise signal. Two different aspects of the loudness calculation procedure should be mentioned atthis point and considered for the analyses in the end of this pa
38、per. First, two neighboring tones, which occurwithin a distance lower than a critical frequency group bandwidth, contribute to the same specific loudness.Second,tones,whichhavealevellowerthanthemaskingthresholdofadjacentfrequencygroups,contributewith the masking threshold of the adjacent frequency g
39、roup level to the specific loudness.Tonality and prominence ratioTonalityof soundand noisesignals isan importantpsychoacoustic featurethat determinesthe annoyanceoreuphonyofitssignal. Thereby,tonalitydescribesthedominanceofsingle tonesor narrow-bandwidthnoisesincomparisontotheremainingnoiselevelofth
40、esignal. Dependingontheinterestofthereceptorinthesoundevent increasing tonality can lead to increased euphony or annoyance. The perception of tonality is fre-quency-dependentandthemaximumofperceptioncanbeobservedatafrequencyof700Hz. Dependingonthe field of applicationthere areseveral standardsand de
41、finitionsof tonality. Thetonality evaluationmethod,proposed by Terhard 16 and Aures 17, gives a time-dependent single value for the airborne noise signal.This evaluation is based on short-time frequency spectra that are evaluated regarding prominent tones andnarrow-bandwidthwithincriticalbandwidths1
42、8. Foreachoftheprominentcomponentsatonalpenaltylevelis calculated against the remaining noise. The single tonality value is finally determined with a non-linearfunction of all these penalty values.In comparison to the summarizing evaluation with the previously described method there are severalappro
43、aches that describe the perception of tonality in a spectrally resolved way. These approaches use asimilarmethodtoidentifytheexistenceofprominenttonesand narrow-bandwidthnoises asthe approachbyTerhardandAures. Buttheygivespectralprominenceratiosortone-to-noiseratiosthatallowevaluatingtherelevance of
44、 frequency regions for the tonal perception of the signal. The evaluation method in DIN 4568119 is a most generalized approach for this objective. This approach can similarly found in standard ISO/IEC 61400-11 20 for wind turbines and in ECMA 74-Annex D 21 for information technology and telecom-muni
45、cations equipment. Generally it can be stated that all tonality evaluation methods have in common thatthey determine a level surplus, respectively a penalty level, of specific tones within a critical bandwidth incomparison to the remaining noise.6 12FTM09SharpnessThepsychoacousticparametersharpnessd
46、escribesthenoiseperceptionthatcorrelateswiththedominanceof high frequencies. With increasing sharpness a noise is rated more aggressive and unpleasant. Thesharpness rating depends on the loudness rating. The unit for sharpness rating is acum. There are severalapproaches for calculating sharpness rat
47、ing. All of them use the specific loudness distribution of the noisesignal. ThestandardDIN 4569222describesamethodforthecalculationofsharpnessbyusingthespecificloudness distribution N(z) according to DIN 45631 15. This specificloudness isweighted bya functiong(z)that puts progressively emphasis on h
48、igher frequency. By normalizing the weighted specific loudnessdistribution to the entire loudness of the signal a weighted frequency center of loudness is determined. Theresulting value defines the sharpness S 14.S = 0.11z=24 Barkz=0N(z) g(z) zBark dzz=24 Barkz=0N(z) dzacum(1)RoughnessSignals,especi
49、allysoundandnoisesignals,areoftenmodulatedinamplitudeorfrequencyinordertotransmitinformation. Spoken language is a typical example for this phenomenon. But when the receptor is notinterested in the information, the modulation can lead to high annoyance, since it attracts attention. Hence,modulation is an important psychoacoustic aspect. When the sound signal is modulated by frequenciesbelow20 Hz,humanperceptioncanfollowenvelopefluctuations18. Above20 Hzthefluctuationtransformsto a perception of rough
