資源描述：
《獨(dú)立分量分析在盲信號(hào)分離中的應(yīng)用研究》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、重慶大學(xué)碩士學(xué)位論文中文摘要摘要近年來，信號(hào)處理的理論與方法獲得了迅速發(fā)展。事實(shí)證明，信號(hào)處理是推動(dòng)眾學(xué)科發(fā)展的一個(gè)重要基石。獨(dú)立分量分析技術(shù)(IndependentComponentAnalysis,簡稱ICA)是信號(hào)處理領(lǐng)域發(fā)展較晚的一種理論與方法，已迅速成為該領(lǐng)域內(nèi)重要的組成部分，且其發(fā)展逐漸趨向成熟化與系統(tǒng)化。本論文對(duì)獨(dú)立分量分析對(duì)于盲信號(hào)分離的應(yīng)用進(jìn)行了研究，主要工作如下：(1)搜集、整理、總結(jié)了國內(nèi)外獨(dú)立分量分析方面的成果和進(jìn)展，介紹了獨(dú)立分量分析的基本理論，并著重討論了幾種常用的獨(dú)立分量分析算法及其特點(diǎn)。(2)介紹了盲
2、信號(hào)分離的基本理論，并在最小互信息盲分離算法的基礎(chǔ)上提出了盲信號(hào)分離的基于模糊神經(jīng)網(wǎng)絡(luò)的獨(dú)立分量分析方法。(3)論文對(duì)基于模糊神經(jīng)網(wǎng)絡(luò)的ICA對(duì)于盲信號(hào)分離的應(yīng)用進(jìn)行了探討，制定了可行的統(tǒng)計(jì)獨(dú)立準(zhǔn)則、邊緣熵和激活函數(shù)，詳細(xì)介紹了模糊神經(jīng)網(wǎng)絡(luò)的設(shè)計(jì)過程，并對(duì)算法的穩(wěn)定性和收斂性進(jìn)行了分析，通過計(jì)算機(jī)實(shí)驗(yàn)證實(shí)了該ICA算法對(duì)于盲信號(hào)分離的良好的性能，指標(biāo)顯示代表信號(hào)特征的信號(hào)的波形和頻率基本與源信號(hào)保持一致達(dá)到了較好的分離效果。與同類算法的性能比較表明該算法分離性能較好。關(guān)鍵詞：獨(dú)立分量分析，盲信號(hào)分離，模糊神經(jīng)網(wǎng)絡(luò)，多道信號(hào)，熵，互信
3、息I重慶大學(xué)碩士學(xué)位論文英文摘要ABSTRACTRecently,theoriesandmethodsofsignalprocessingareobtainedtodevelopquickly.Infact,signalprocessingisanimportantfoundationstoneforallkindsofsubjects'developing.Independentcomponentanalysis(ICA)technologydevelopingisalatertheoryormethodinfieldsofsig
4、nalprocessing,butitwasimportantthatithasrapidlybecomeapartofconstitutionofsignalprocessingfields，anditsdevelopingtendsgraduallytomaturityandsystematization.Inthispaper,thetheoryandapplicationofindependentcomponentanalysishasbeenstudied.Thepaperconsistsoffollowingparts:
5、1.First,thepaperreviewssystematicallythepresentresearchsituationofindependentcomponentanalysisintheworld.Thebasicprinciplesandconceptsofindependentcomponentanalysisandsomealgorithmsareintroduced.2.Second,aminimummutualinformationalgorithmbasedrecurrentneuralnetworksare
6、givenafterintroducedthebasictheoryaboutBlindSourceSeparation.Themainprinciple,derivationhasbeenpresented.3.Applicationresearchofblindsignalseparationbyusingindependentcomponentanalysishasbeenstudiedinthispaper.Thefeasibleindependencerule,brimentropyandactivationfunctio
7、nhavebeenestablished.Thedesignoffuzzyneuralnetworkshasbeenintroducedparticularly.Andtheperformanceofthealgorithmhasbeenillustratedbycomputationsimulationexperiment.Theresultdisplaytheperformanceofthisalgorithmisgood.Keywords:Independentcomponentanalysis,Blindsourcesepa
8、ration,Fuzzyneuralnetworks,Entropy,mutualinformation,II重慶大學(xué)碩士學(xué)位論文1緒論1緒論引言近幾年，盲源分離(BlindSourceSeparation:BSS)已經(jīng)成為信號(hào)處理領(lǐng)
