小波阈值神经网络在信号去噪及预测中的应用

《小波阈值神经网络在信号去噪及预测中的应用》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、25ज3௹॥ᇅ৘ંაႋႨVol.25No.32008୍6ᄅControlTheory&ApplicationsJun.2008໓໓ᅣᅣᅣщщщ:1000−8152(2008)03−0485−07ཬཬѯѯѯᚐᚐᚐᆴᆴᆴപപപࣜࣜࣜຩຩຩ઎઎઎ᄝᄝᄝྐྐྐಀಀಀᄮᄮᄮࠣࠣࠣყყყҩҩҩᇏᇏᇏႋႋႋႨႨᖳၴې1,ໍდ2,෤Џ3(1.Кࣘࢌ๙ն࿐ྐ༏॓࿐࿹࣮෮,Кࣘ100044;2.໳॓࠯ն࿐ྐ༏۽ӱა॓࠯࿐ჽ,޵К໳430081;3.޿ᇏ॓࠯ն࿐॥ᇅ॓࿐ა໳К޵,༢ӱ۽430074)ᅋေ:ิԛਔ၂ᇕཬѯᚐᆴപࣜຩ઎ଆ྘(wave

2、letthresholdneuralnetwork,WTNN),ؓކቔൔࢤ൬ূղྐࣉྛಀᄮބყҩ.ᆃᇕຩ઎ଆ྘ϜཬѯቋႪᚐᆴಀᄮఖࡆപࣜຩ઎ᇏ,ؓջᄮྐऎႵཬѯቋႪᚐᆴಀᄮބყҩٳྛࣉمෘ਀࿞ؓݖ๙,Ցః.ࡥࣚӱщ൐,مෘ਀࿞߄ࡥॖ,ܒᇗҪֆቔඔ༢ѯཬؓ.ିۿ༅,ԛਔቋႪᚐᆴࠣಃᆴטᆜ܄ൔ.ቋު๙ݖؓཌྟט௔ྐಀᄮࠣభཟ၂ყҩൌဒࢲݔॖၛुԛ,ຩ઎ൻೆٳљູջႵۚථϢᄮല~ۚථջཋᄮല~ೋ০ᄮലཌྟט௔ྐൈWTNNࢲݔनႪႿ০ႨDonohoᚐᆴࣉྛಀᄮުᄜყҩࢲݔ.ܱ࡯Ս:ቋႪᚐᆴ;പࣜຩ઎;ყҩ;ྐಀᄮᇏ๭োٳ:TP27

3、3໓ངѓ്઒:ATheapplicationofwaveletthresholdneuralnetworkinthede-noisingandpredictionCENYi-gang1,WEIYu2,SUNDe-bao3(1.InstituteofInformationScience,BeijingJiaotongUniversity,Beijing100044,China;2.SchoolofInformationEngineering&Technology,WuhanUniversityofScience&Techno

4、logy,WuhanHubei430081,China;3.DepartmentofControlScience&Engineering,HuazhongUniversityofScienceandTechnology,WuhanHubei430074,China)Abstract:Awaveletthresholdneuralnetwork(WTNN)modelisproposedfordenoisingandpredictionofcooperativelyreceivedradarsignals.ThisWTNNin

5、corporatesawaveletdenoisinglayerwithoptimalwaveletthresholdsintotheneuralnetwork,forsignaldenoisingandpredicting.Thetrainingalgorithmissimplifiedbythesingle-layerreconstructionofwaveletcoefficients,leadingtoacompactprogramming.Byanalyzingthetrainingalgorithm,wederiv

6、ethetuningformulasforsearchingoptimalthresholdsandnetworkweights.Theresultsofdenoisingandone-stepaheadpredictionforalinearfrequencymodulationsignalwithwhiteGaussnoise,Gaussband-limitednoiseorRayleighnoiseshowthattheWTNNperformsmuchbetterthanthemethodofDonoho-thres

7、holdfordenoisingandprediction.Keywords:optimalthreshold;neuralnetwork;prediction;signalde-noising1ႄႄ࿽࿽(Introduction)τjູjҪᚐᆴ,dj,kູjҪ༥ࢫ༢ඔ,d˜j,kູᚐᆴགྷൌൗࢸᇏᄮല҂ॖх૧,ྐಀᄮ൞ଢభԩ৘ުjҪ༥ࢫ༢ඔ.√၂۱࿹࣮ಣ.ಀᄮ༵ေࣉྛ༢๤ࡹଆ,ᆃିࠞնࣜDonoho௴ൡᚐᆴູτj=ˆσj2lnN,֌ھᚐิശಀᄮིݔ[1].Donohoิԛਔ٤ཌྟཬѯᚐᆴಀᆴᆺ൞ᄮല၂۱ܙ࠹.൳Мࣟᄮല႕ཙ

8、ࢠն,ၹՎᄮ[2,3],ܼمٚᆴᚐႪቋࠆᅳ࿙ႋᆌ؟ᄮಀᆴᚐѯཬ,భଢ.]5,4]Ⴈႋٗ.Ч໓ิԛ၂ᇕྍھ,é)WTNN(઎ຩࣜപᆴᚐѯཬç:઎ຩࣜപק၂҂ᆴᚐohonoD,ലᄮ೤Ⴕؓ,ലᄮϢථۚႿؓຩ઎ିࢠಀᄮིݔ.ᚐᆴಀᄮЇও႗ᚐᆴაೈᚐࠢӮਔཬѯᚐᆴಀᄮ,൐ಀᄮაྐყҩ൐Ⴈ๝၂ᆴ,ఃᇏೈᚐᆴٚمູ༂