Full metadata record
DC FieldValueLanguage
dc.contributor.author翁文祥en_US
dc.contributor.authorWen-Shiung Wengen_US
dc.contributor.author鄧清政en_US
dc.contributor.authorChing-Cheng Tengen_US
dc.date.accessioned2014-12-12T02:24:07Z-
dc.date.available2014-12-12T02:24:07Z-
dc.date.issued1999en_US
dc.identifier.urihttp://140.113.39.130/cdrfb3/record/nctu/#NT880591010en_US
dc.identifier.urihttp://hdl.handle.net/11536/66240-
dc.description.abstract本論文的目的在於嘗試建構一個以傅立葉級數為基底的新的函數模擬網路架構(FBN),以達到模擬週期性函數的目的,亦或藉由架設一平行週期偵測網路補強原本以高斯函數為基底的函數模擬網路(FNN)模擬週期性函數能力的不足。本論文所達到的結果是:FBN有不錯的模擬週期性函數的能力、改善補強後的 FNN有模擬週期性函數的能力以及改變FBN基底後的函數模擬網路,可用來設計任意規格FIR濾波器。zh_TW
dc.description.abstractIn this thesis, we present a new network (Fourier Based Network, (FBN)) to solve the approximation problem of periodic functions. The basis of the FBN are sine and cosine functions, hence we call it "Fourier Based Network". Because of this characteristic, the Fourier series expansion of a given function can be obtained by the FBN. Moreover, the FBN also can be applied to estimate the signal period from the given data and design the coefficient of the FIR filter that has the specification which the system required. To illustrate the effectiveness of our approach, the Fourier base Network is applied to simulate many signals including 1D and 2D whether its period is known or not. At the end of this thesis, we compensate the ability of simulating periodical function of Fuzzy Neural Network(FNN) by adding a parallel period-searching network. 1.1 研究動機…………………………………………………………………….1 1.2 論文架構…………………………………………………………………….3 第二章 網路架構描述及理論推導…………………………………………………..4 2.1 FBN網路架構…………………..…………………………………………...4 2.2誤差回傳推導………………………………………………………………..4 2.3收斂性質推導………………………………………………………………..6 2.4 一維FBN架構的推廣……………………..……………………………….8 2.5 FNN簡介及和FBN之間的比較…..…………………………………….…9 2.5.1 FNN簡介……………………………………………………..……….9 2.5.2 FNN和FBN之間的比較………………………………………………11 第三章 範例及模擬結果……………………………………………………………12 3.1 已知週期訊號模擬………………………………………………………...13 (一) 無雜訊訊號……………………………………. ……………………...13 a. 用一維FBN模擬已知週期無雜訊一維訊號 ………………………13 b. 二維FBN模擬已知週期無雜訊的二維訊號…………..…………….15 (二) 有雜訊訊號…………………………………………….………………17 a. 一維FBN模擬一維純雜訊訊號……….…………………………….17 b. 一維FBN模擬帶有雜訊已知週期的一維訊號………………………18 c. 二維FBN模擬二維純雜訊訊號………………………………..…….21 d.以二維FBN模擬已知週期參雜雜訊的二維訊號……………….……22 3.2 未知週期訊號模擬………………………………………………………...24 (一) 無雜訊訊號…………………………………………………………….24 a. 以一維FBN模擬未知週期無雜訊一維訊號(訊號長度大於完整週期)…………………………………………………………………..24 b. 以一維FBN模擬未知週期無雜訊一維訊號(訊號長度小於完整週期)…………………………………………………………………..26 c. 用二維FBN模擬二維未知週期無雜訊訊號,其兩軸訊號長度均大於完整週期……………………………………………………………..27 d. 用二維FBN模擬二維未知週期訊號,其兩軸訊號長度均小於完整週期……………………………………………………………………..29 (二) 有雜訊訊號…………………………………………………………….31 a. 用含頻率變數的FBN模擬一維訊號,訊號長度大於完整週期………………………………………………….………………….31 b. 用含頻率變數的FBN模擬一維含雜訊未知週期訊號,訊號長度小於完整週期……………………………………………………..………33 c. 用含頻率變數的二維FBN模擬二維未知週期且參雜雜訊之二維訊號,訊號長度大於完整週期………………………………………..34 d. 用含頻率變數的二維FBN模擬二維未知週期且參雜雜訊之二維訊號,訊號長度小於完整週期………………………………………..36 第四章FBN 應用於FIR濾波器的設計…………………………………………..39 4.1 FBN應用於一維FIR濾波器的設計……………………………………..39 4.2 FBN應用於二維FIR濾波器的設計………………………………………44 第五章 改良後之 FNN 的模擬結果……………………………………………...46 第六章 結論與展望…………………………………………………………………49 參考文獻 ……………………………………………………………………………50en_US
dc.language.isozh_TWen_US
dc.subject傅立葉級數zh_TW
dc.subject函數模擬網路zh_TW
dc.subjectFIR濾波器zh_TW
dc.subject週期性函數zh_TW
dc.subject高斯函數zh_TW
dc.subject類神經網路zh_TW
dc.subject模糊zh_TW
dc.subject誤差回傳zh_TW
dc.subjectfourier serieen_US
dc.subjectfunction approximation networken_US
dc.subjectFIR filteren_US
dc.subjectperiodical functionen_US
dc.subjectGaussian functionen_US
dc.subjectneural networken_US
dc.subjectfuzzyen_US
dc.subjecterror propagationen_US
dc.title以傅立葉級數為基底的函數近似網路及其應用zh_TW
dc.titleFunction Approximation Network and its Application Based on Fourier Seriesen_US
dc.typeThesisen_US
dc.contributor.department電控工程研究所zh_TW
Appears in Collections:Thesis