流形算法及其水電機組噪聲源檢測.pdf

1、浙江工業(yè)大學(xué)碩士學(xué)位論文流形算法及其水電機組噪聲源檢測姓名：陳旻昊申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用技術(shù)指導(dǎo)教師：王萬良鄭建煒20121016浙江工業(yè)大學(xué)碩士學(xué)位論文ManifoldLearningandItsApplicationinTheNoiseSourceDetectionofHydroelectricGeneratingUnitABSTRACTSmallhydropowerisakindofclearandreproducib

2、lepowerandisthekeystrategicdirectionofthenationalenergydevelopmentButatpresentmosthydroelectricgeneratingunitsdetecttheabnormalnoiseandfaultbyartificialjudgementThiswayhaslowspeedandwillinfluencetheaccuracyofjudgementSot

3、hatitisnecessarytoconsiderthemachinelearningmethodManifoldlearninghastheabilityoffindingtheinnerstructureofthedatasetItisvaluabletesearchthewaytoapplythemanifoldlearningalgorithmtothehydroelectricgeneratingunitnoisesourc

4、edetectionThisarticlehasanalysedthecharacteristicsofthehydroelectricgeneratingunitnoisesignalandtheproblemswhenapplyingthemanifoldlearningtothehydroelectricgeneratingunitnoisesourcedetection，andthendesignedthecorrespondi

5、ngmanifoldlearningalgorithmThemaincontributionsoftheworkaleasfollow：1)Analyzethereasonsofabnormalvibrationofthehydroelectricgeneratingunitandthecharacteristicsofthenoisesignalandintroducethevibrationsignaldenoisingmethcd

6、indetails2)Designanincrementalwithinclasslocalitypreservingdimensionreductionalgorithm，IWDAThisalgorithmisdesignedtosolvethelinearproblemandhastheabilityoftheincrementalcomputationandCanpreservethelocalspacestructureThis

7、algorittuniseffectiveonthedatawhichismultimodeoroverlapping3)DesignallincrementalkerneldiscriminationanalysismethodviaQRDecomposition，IKDR／QRThisalgorithmissuitableforcomplexnolineardevicelikehydroelectricgeneratingunitb

8、ycombiningIWDAwithkernelmethodBecausetheabnormalvibrationnoiseishardtogetintheearlystageofthenoisedetection，theonlinelearningisneededIKDPUQRCandecreasethetimeandspaceofcreatingthekernelmatrixduringtheprocessofonlinelearn

9、ing4)ApplyIKDR／QRtohydroelectricgeneratingunitnoisesourceonlinedetectionThenoisesourceonlinedetectionexperimentprovedtheeffectivenessofIKDR／QRKeyWords：hydroelectricgeneratingunit，noisesourceidentification，onlineleanfing，