自適應(yīng)Volterra濾波器的遞歸結(jié)構(gòu)和類遞歸結(jié)構(gòu)及其算法和應(yīng)用
RECURSIVE STRUCTURE AND QUASI-RECURSIVE STRUCTURE OF ADAPTIVE VOLTERRA FILTERS AND THEIR ALGORITHMS AND APPLICATIONS
-
摘要: 本文參考自適應(yīng)IIR濾波器理論,提出了自適應(yīng)Volterra濾波器(AVF)的遞歸結(jié)構(gòu)和類遞歸結(jié)構(gòu),討論了其特點(diǎn)和應(yīng)用范圍.遞歸結(jié)構(gòu)的引入可顯著減少AVF的參數(shù)和計算量.本文還給出了類遞歸結(jié)構(gòu)AVF的在線辨識算法和在非線性系統(tǒng)辨識中的應(yīng)用;給出了遞歸結(jié)構(gòu)AVF的濾波算法和在非線性相關(guān)噪聲抵消中的應(yīng)用.在仿真實(shí)驗(yàn)中,將上述算法與多層感知器和非遞歸結(jié)構(gòu)AVF做了對比.結(jié)果表明,本文算法住性能和計算量上均有明顯優(yōu)勢。Abstract: In reference of the theory of adaptive IIR filters, the paper puts forward the recursive structure and quasi-recursive structure of Adaptive Volterra Filters(AVF), and discusses their characteristics and areas of applications. The introduction of recursive structure can remarkably reduce the parameters and computational cost of AVF. The on-line identification algorithm of quasi-recursive structure AVF with its application in non-linear system identification and the filtering algorithm of recursive structure AVF with its application in non-linear correlated noise cancellation are also given. In simulations, the above algorithms are compared with multi-layered perceptron and non-recursive AVF. The results show the algorithms of the paper have obvious advantages both in performance and in computational cost.
-
Sandberg I W. On Volterra expansions for time-varying nonlinear systems. IEEE Trans. on Circuit[2]and Syst.,1983, CAS-30.(2): 61-67.[3]羅發(fā)龍,李衍達(dá).神經(jīng)網(wǎng)絡(luò)與信號處理.北京:電子工業(yè)出版,1993.[4]Sbynk J J. Adaptive IIR filtering. IEEE Signal Processing Mag.,1989,6(2): 4-21.[5]Piche S P. Steepest descent algorithms for neural controllers and filters. IEEE Trans. on Neural Networks, 1994, NN-5:198-212. -
計量
- 文章訪問數(shù): 2319
- HTML全文瀏覽量: 153
- PDF下載量: 513
- 被引次數(shù): 0
下載:
