摘 要:自適應(yīng)信息處理的算法中的RLS算法在信號處理等方面已經(jīng)得到了大量的應(yīng)用。首先簡要介紹了RLS算法,然后通過對RLS算法中的遺傳因子的研究與分析,提出了一種新的變遺忘因子算法,通過修正函數(shù)對遺傳因子進(jìn)行修正,實(shí)現(xiàn)了此算法的優(yōu)勢,最后對該算法做了仿真試驗(yàn)。試驗(yàn)證明,此算法收斂速度和跟蹤效果遠(yuǎn)好于普通RLS算法,并且系統(tǒng)穩(wěn)定,具有較強(qiáng)的應(yīng)用價(jià)值。
關(guān)鍵詞:遺忘因子;RLS算法;均方誤差;修正函數(shù)
中圖分類號:TN92 文獻(xiàn)標(biāo)識碼:B 文章編號:1004373X(2008)1704503
Analysis and Simulation of a Variable Forgetting Factor RLS Algorithm
LI Qianru1,WANG Yuding1,ZHANG Xiaofang2
(1.Xi′an Communication Institute,Xi′an,710106,China;
2.School of Information and Control Engineering,Xi′an University of Architecture and Technology,Xi′an,710055,China)
[JP2]Abstract:The RLS algorithm in the adaptive information processing algorithm is widely used in signal processing.In this paper,RLS algorithm is firstly introduced,with the help of research and analysis to forgetting factor of RLS algorithm,a new kind of variable forgetting factor algorithm is put forward.This algorithm is superior to common RLS algorithm with the help of adjusting the forgetting factor by the adjustable-function,the computer experiments for the algorithm are done.The experiments indicate that the convergence speed and tracking effort of this new algorithm are far better than common RLS algorithm′s,the system is steady and has important application value.
Keywords:forgetting factor;RLS algorithm;mean-square error;adjustable function
1 引 言
自適應(yīng)信息處理的算法、方案繁多,究其實(shí)質(zhì)可歸納為遵循最小均方誤差(Least Mean Square,LMS)準(zhǔn)則及最小二乘(Least Square,LS)準(zhǔn)則兩大類,其他算法大多是這兩種算法的演進(jìn)。普通的LMS算法跟蹤能力強(qiáng),但收斂速度不是很快;而普通的遞推RLS算法跟蹤能力又有待提高,所以改進(jìn)普通的遞推RLS算法可以更多地實(shí)現(xiàn)性能優(yōu)異的濾波器。遞推RLS算法中的遺忘因子對系統(tǒng)的性能起到了非常關(guān)鍵的作用。調(diào)整遺忘因子推導(dǎo)出來的算法,具有收斂速度快,跟蹤能力強(qiáng)的優(yōu)點(diǎn),仿真結(jié)果表明改進(jìn)以后的算法具有較小的參數(shù)估計(jì)誤差,數(shù)值穩(wěn)定性好。
2 普通RLS算法
最小二乘濾波大約是1795年高斯在星體運(yùn)動(dòng)軌道預(yù)測研究中提出的。它的基本結(jié)果有兩種形式,一種是經(jīng)典的一次完成算法,另一種是現(xiàn)代的遞推算法(Recursive of Least Square,RLS)。經(jīng)典算法在理論研究中更為方便,而RLS算法適合于計(jì)算機(jī)處理。文獻(xiàn)[1]在推導(dǎo)RLS算法的時(shí)候,為了簡單起見,將遺忘因子λ=1處理,文獻(xiàn)[2]對λ作為參數(shù)推導(dǎo)了遞推RLS算法。
注:本文中所涉及到的圖表、注解、公式等內(nèi)容請以PDF格式閱讀原文