Detection of AV-fistula abnormalities in hemodialysis patients

100 Pages·2012·5.81 MB·English

DETECTION AND RATINGS
OF AUTOMATIVE IMPULSIVE
EVENTS
Master Thesis
Martin Kynde
Spring 2012

InstituteofElectronicSystems
FrederikBajersVej7B5
9220AalborgØ
Phone[phone]
Fax[phone]
http://www.sict.aau.dk
Projecttitle:
Detectionandratingsofautomative
impulsiveevents
Abstract:
Semestertheme: Thedetectionandratingsofimpulsescanbeused
in wide fields within the industry. An engineer
MasterThesis
working with detecting impulses and reducing
unwanted sounds in consumer products, most
prominentjobistopreventallannoyingsounds
Projectperiod: as impulses to be perceived by the costumer.
MasterPrograminAcoustics Soundqualityisapartofthetotalproductquality.
4th. semester,Spring2012
In the area of NVH engineering they are often
dealtwithdetectingandeliminateBuzz,Squeak
andRattle(BSR)eventsinvehicles.
Projectgroup:
Manyattemptshasbeendoneinordertoauto-
12gr1061 maticdetectBSReventsandoftenthedetection
threshold is depending on an absolute level
determined by a subjective evaluation or set
Groupmember
manuallybytheauthors.
MartinKynde
In this project an algorithm independent of
Supervisor: absolute level has been chosen to calculate the
amountofimpulsivenessofanacousticsignal.
ChristianSejerPedersen
Synthetically generated samples consisting of
impulsesmixedwithnoisewereusedasstimuli
to the proposed algorithm. Concurrently the
Secondarysupervisor
same set of samples has been evaluated in a
Woo-KeunSong,Brüel&Kjær subjective experiment. Existing literature does
notdescribeasimilarstudyofimpulseswiththe
mentionedparameters,conductedinasubjective
experiment. Thatwasthemotivationtoexplore
No.ofprintedcopies: 5 thedescribedscenario.
The goal of the project was to reveal if any
No.ofpages: 53
correlation existed between the objective and
subjectiveevaluationofthesamples.
No.ofappendixpages: 27
Nosignificantcorrelationwasfoundbetweenthe
two. Futureworksuggesttotunethesettingsin
Totalno.ofpages: 87
thealgorithmtoevaluateifanimprovementcan
beachievedthatmatchthesubjectiveresultswith
ahigherdegreeofcorrelation.
Completed:
May31,2012
Thecontentsofthisreportarefreelyaccessible,howeverpublication(withsourcereferences)
isonlyalloweduponagreementwiththeauthor.
iv

Contents
Contents
Readingguide i
1 Introductionandproblemanalysis 3
2 SystemOverview 9
3 ImpulsiveSounds 11
3.1 Impulses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.4 Post-processingofsamples . . . . . . . . . . . . . . . . . . . . 14
3.4.1 Inversefiltering . . . . . . . . . . . . . . . . . . . . . . 15
3.4.2 Leveladjustment . . . . . . . . . . . . . . . . . . . . . 16
4 Methodsforevaluatingthesamples 17
4.1 Subjectiveevaluation . . . . . . . . . . . . . . . . . . . . . . . 17
4.2 Objectiveevaluation . . . . . . . . . . . . . . . . . . . . . . . . 18
4.2.1 RelativeApproach . . . . . . . . . . . . . . . . . . . . 18
4.2.2 Calculationofimpulsiveness . . . . . . . . . . . . . . 20
4.2.3 InputmodelstotheRelativeApproachalgorithm . . 21
4.2.4 Testofalgorithms . . . . . . . . . . . . . . . . . . . . . 26
5 Experiment 29
5.1 Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.2 Experimentalsetup . . . . . . . . . . . . . . . . . . . . . . . . 29
5.3 Experimentalprocedure . . . . . . . . . . . . . . . . . . . . . 31
6 Results 37
6.1 Subjectiveresults . . . . . . . . . . . . . . . . . . . . . . . . . 37
6.2 Objectiveresults . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.3 Correlation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
7 Discussion 51
7.1 Correlationbetweenobjectiveandsubjectiveresults . . . . 52
8 Conclusions 55
A Samples 59
B Experimentformailites 63
C Additionalresults 69
CONTENTS
D Measurement of BeyerDynamic DT990 Pro headphones + Ear
transferfunctions 75
E EnclosedDVDcontents 85
Bibliography 87
Reading guide
ABSTRACT
Thedetectionandratingsofimpulsescanbeusedinwidefieldswithin
theindustry. Anengineerworkingwithdetectingimpulsesandreducing
unwanted sounds in consumer products, most prominent job is to
preventallannoyingsoundsasimpulsestobeperceivedbythecostumer.
Soundqualityisapartofthetotalproductquality.
In the area of NVH engineering they are often dealt with detecting and
eliminateBuzz,SqueakandRattle(BSR)eventsinvehicles.
Many attempts has been done in order to automatic detect BSR events
and often the detection threshold is depending on an absolute level
determinedbyasubjectiveevaluationorsetmanuallybytheauthors.
In this project an algorithm independent of absolute level has been
chosentocalculatetheamountofimpulsivenessofanacousticsignal.
Synthetically generated samples consisting of impulses mixed with
noise were used as stimuli to the proposed algorithm. Concurrently
the same set of samples has been evaluated in a subjective experiment.
Existingliteraturedoesnotdescribeasimilarstudyofimpulseswiththe
mentionedparameters,conductedinasubjectiveexperiment. Thatwas
themotivationtoexplorethedescribedscenario.
The goal of the project was to reveal if any correlation existed between
theobjectiveandsubjectiveevaluationofthesamples.
Nosignificantcorrelationwasfoundbetweenthetwo. Futureworksug-
gest to tune the settings in the algorithm to evaluate if an improvement
canbeachievedthatmatchthesubjectiveresultswithahigherdegreeof
correlation.
Thereportisstructuredinthefollowingway.
CHAPTER 1 Introductionandproblemanalysis
ContainsIntroductionandproblemanalysis. Anoverviewoftheim-
portanceofdetectingimpulsesintheindustryisdescribed. Existing
methodsfordetectingimpulsesislistedandthemainproblemofthis
projectisstated.
CHAPTER 2 Systemoverview
Contains a block-diagram of the main features of the project and
linksittothecorrespondingchapters.
i
ii READINGGUIDE
CHAPTER 3 ImpulsiveSounds
Contains a description of the different parameters involved in de-
signing the samples to the objective and subjective evaluation and
thetechnicalaspectsofachievingaflatfrequencyresponseandthe
correctlevelsconcerningthesubjectiveexperiment.
CHAPTER 4 Methodsforevaluationthesamples
Contains a description of the subjective and objective parameters
usedtoevaluatethesamples.
CHAPTER 5 Experiment
Contains information about the psychoacoustic experiment con-
ductedinordertoobtainsubjectiveratingsofthesamples.
CHAPTER 6 Results
Containsresultsfrom; Thesubjectiveexperiment, theobjectivecal-
culationofimpulsivenessandthecorrelationbetweenthem.
CHAPTER 7 Discussion
Contains discussion about the issues that were encountered when
analysing the subjective results and the correlation between objec-
tiveparameterswithsubjectivescales.
CHAPTER 8 Conclusion
Containsasummaryofwhathasbeendoneintheprojectandgives
ananswerofifthemaingoalhasbeenfulfilled.
Appendices:
APPENDIX A Samples
Containsadiscussionaboutthechosensettingsforeachparameter
in the design of samples. The number of samples are constrained
by time limitations stated in the experiment chapter. The effective
testingtimeisestimated.
APPENDIX B Experimentformalities
ThefirstpartoftheappendixcontainsaDanishInstructionlistand
atranslatedEnglishversion. Thesecondpartcontainsanswersfrom
thesubjectscollectedinthebreaksoftheexperiment.
APPENDIX C Additionalresults
Contains additional results from the subjective analyse. Important
conclusionsaredrawnintermsofannoyanceandimpulsiveness.
APPENDIX D MeasurementofBeyerDynamicDT990Pro
headphones+Eartransferfunctions
Contains a measurement journal of the transfer function measure-
ment of BeyerDynamic DT990 Pro used for the subjective experi-
ment.
APPENDIX E EnclosedDVDcontents
ContainsalistofmaterialincludedontheenclosedDVD.