摘要
Apre-filtercombinedwiththresholdself-learningwaveletalgorithmisproposedforhydraulicpressuresignalsdenoising.Thedenoisingthresholdisself-learntinthesteadyflowstate,andthenmodifiedunderagivenlimittomakethemeansquareerrorsbetweenreconstructionsignalsanddesirableoutputsminimum,sothecorrespondingoptimaldenoisingthresholdinasingleoperatingcasecanbeobtained.Theseoptimalthresholdsareusedforthewholesignaldenoisingandaredifferentinvariouscases.Simulationresultsandcomparativestudiesshowthatthepresentapproachhasanobviouseffectofnoisesuppressionandissuperiortothoseoftraditionalwaveletalgorithmsandback-propagationneuralnetworks.Italsoprovidestheprecisedataforthenextstepofpipelineleakdetectionusingtransienttechnique.
出版日期
2008年04月14日(中国期刊网平台首次上网日期,不代表论文的发表时间)