Location:
Search - LMS E
Search list
Description: 基于Volterra滤波器混沌时间序列多步预测
作者:陆振波,海军工程大学
欢迎同行来信交流与合作,更多文章与程序下载请访问我的个人主页
电子邮件:luzhenbo@sina.com
个人主页:luzhenbo.88uu.com.cn
参考文献:
1、张家树.混沌时间序列的Volterra自适应预测.物理学报.2000.03
2、Scott C.Douglas, Teresa H.-Y. Meng, Normalized Data Nonlinearities for LMS Adaptation. IEEE Trans.Sign.Proc. Vol.42 1994
文件说明:
1、original_MultiStepPred_main.m 程序主文件,直接运行此文件即可
2、original_train.m 训练函数
3、original_test.m 测试函数
4、LorenzData.dll 产生Lorenz离散序列
5、normalize_1.m 归一化
6、PhaSpaRecon.m 相空间重构
7、PhaSpa2VoltCoef.dll 构造 Volterra 自适应 FIR 滤波器的输入信号矢量 Un
8、TrainTestSample_2.m 将特征矩阵前 train_num 个为训练样本,其余为测试样本
9、FIR_NLMS.dll NLMS自适应算法-based Volterra filters chaotic time series multi-step forecast Author : bo, the Navy Engineering from the University of peer welcome exchanges and cooperation, more and download articles please visit my personal web page e-mail : luzhenbo@sina.com WEBSITE : luzhenbo.88uu.com.cn References : 1, and Zhang Shu. the chaotic time series Volterra adaptive prediction. physics reported .2000.03 2, Scott C. Douglas, H.-Y. Teresa Meng, Normalized Data for LMS Adaptation Nonlinearities. IEEE Trans.Sign.Proc . Timing 1994 document : 1, original_MultiStepPred_main.m procedures master file directly run this document can be 2, 3 original_train.m training function, the function tests original_test.m 4, LorenzData.dll have Lorenz five discrete sequence, a normalize_1.m naturalization of six, PhaSpaRecon.m
Platform: |
Size: 11847 |
Author: 陆振波 |
Hits:
Description: Some algorithms of variable step size LMS adaptive filtering are studied.The VS—LMS algorithm is improved.
Another new non-linear function between肛and e(/ t)is established.The theoretic analysis and computer
simulation results show that this algorithm converges more quickly than the origina1.Furthermore,better antinoise
property is exhibited under Low—SNR environment than the original one.-variable step size of a LMS daptive filtering are studied. The VS-LMS algorithm is improved. Another new non-linear function between anus and e (/ t) is established. The theoretic analysis and computer simulatio n results show that this algorithm converges mo 're quickly than the origina1. Furthermore, better antinoise property is exhibited under L ow - SNR environment than the original one.
Platform: |
Size: 3701 |
Author: 上将 |
Hits:
Description: In 1960, R.E. Kalman published his famous paper describing a recursive solution
to the discrete-data linear filtering problem. Since that time, due in large part to advances
in digital computing, the Kalman filter has been the subject of extensive research
and application, particularly in the area of autonomous or assisted
navigation.-In 1960, R. E. Kalman published his famous paper describ ing a recursive solution to the discrete-data l inear filtering problem. Since that time, due in large part to advances in digital computi Vi, the Kalman filter has been the subject of extens ive research and application. particularly in the area of autonomous or assis ted navigation.
Platform: |
Size: 1418 |
Author: 上将 |
Hits:
Description: 若不希望用与估计输入信号矢量有关的相关矩阵来加快LMS算法的收敛速度,那么可用变步长方法来缩短其自适应收敛过程,其中一个主要的方法是归一化LMS算法(NLMS算法),变步长 的更新公式可写成
W(n+1)=w(n)+ e(n)x(n)
=w(n)+ (3.1)
式中, = e(n)x(n)表示滤波权矢量迭代更新的调整量。为了达到快速收敛的目的,必须合适的选择变步长 的值,一个可能策略是尽可能多地减少瞬时平方误差,即用瞬时平方误差作为均方误差的MSE简单估计,这也是LMS算法的基本思想。
Platform: |
Size: 3285 |
Author: 闫丰 |
Hits:
Description: 基于Volterra滤波器混沌时间序列多步预测
作者:陆振波,海军工程大学
欢迎同行来信交流与合作,更多文章与程序下载请访问我的个人主页
电子邮件:luzhenbo@sina.com
个人主页:luzhenbo.88uu.com.cn
参考文献:
1、张家树.混沌时间序列的Volterra自适应预测.物理学报.2000.03
2、Scott C.Douglas, Teresa H.-Y. Meng, Normalized Data Nonlinearities for LMS Adaptation. IEEE Trans.Sign.Proc. Vol.42 1994
文件说明:
1、original_MultiStepPred_main.m 程序主文件,直接运行此文件即可
2、original_train.m 训练函数
3、original_test.m 测试函数
4、LorenzData.dll 产生Lorenz离散序列
5、normalize_1.m 归一化
6、PhaSpaRecon.m 相空间重构
7、PhaSpa2VoltCoef.dll 构造 Volterra 自适应 FIR 滤波器的输入信号矢量 Un
8、TrainTestSample_2.m 将特征矩阵前 train_num 个为训练样本,其余为测试样本
9、FIR_NLMS.dll NLMS自适应算法-based Volterra filters chaotic time series multi-step forecast Author : bo, the Navy Engineering from the University of peer welcome exchanges and cooperation, more and download articles please visit my personal web page e-mail : luzhenbo@sina.com WEBSITE : luzhenbo.88uu.com.cn References : 1, and Zhang Shu. the chaotic time series Volterra adaptive prediction. physics reported .2000.03 2, Scott C. Douglas, H.-Y. Teresa Meng, Normalized Data for LMS Adaptation Nonlinearities. IEEE Trans.Sign.Proc . Timing 1994 document : 1, original_MultiStepPred_main.m procedures master file directly run this document can be 2, 3 original_train.m training function, the function tests original_test.m 4, LorenzData.dll have Lorenz five discrete sequence, a normalize_1.m naturalization of six, PhaSpaRecon.m
Platform: |
Size: 11264 |
Author: 陆振波 |
Hits:
Description: Some algorithms of variable step size LMS adaptive filtering are studied.The VS—LMS algorithm is improved.
Another new non-linear function between肛and e(/ t)is established.The theoretic analysis and computer
simulation results show that this algorithm converges more quickly than the origina1.Furthermore,better antinoise
property is exhibited under Low—SNR environment than the original one.-variable step size of a LMS daptive filtering are studied. The VS-LMS algorithm is improved. Another new non-linear function between anus and e (/ t) is established. The theoretic analysis and computer simulatio n results show that this algorithm converges mo 're quickly than the origina1. Furthermore, better antinoise property is exhibited under L ow- SNR environment than the original one.
Platform: |
Size: 3072 |
Author: 上将 |
Hits:
Description: In 1960, R.E. Kalman published his famous paper describing a recursive solution
to the discrete-data linear filtering problem. Since that time, due in large part to advances
in digital computing, the Kalman filter has been the subject of extensive research
and application, particularly in the area of autonomous or assisted
navigation.-In 1960, R. E. Kalman published his famous paper describ ing a recursive solution to the discrete-data l inear filtering problem. Since that time, due in large part to advances in digital computi Vi, the Kalman filter has been the subject of extens ive research and application. particularly in the area of autonomous or assis ted navigation.
Platform: |
Size: 1024 |
Author: 上将 |
Hits:
Description: 若不希望用与估计输入信号矢量有关的相关矩阵来加快LMS算法的收敛速度,那么可用变步长方法来缩短其自适应收敛过程,其中一个主要的方法是归一化LMS算法(NLMS算法),变步长 的更新公式可写成
W(n+1)=w(n)+ e(n)x(n)
=w(n)+ (3.1)
式中, = e(n)x(n)表示滤波权矢量迭代更新的调整量。为了达到快速收敛的目的,必须合适的选择变步长 的值,一个可能策略是尽可能多地减少瞬时平方误差,即用瞬时平方误差作为均方误差的MSE简单估计,这也是LMS算法的基本思想。
-Want to estimate if the input signal vector and the relevant matrix to speed up the convergence rate of LMS algorithm, then the variable step size method can be used to shorten its adaptive convergence process, one of the main method is normalized LMS algorithm (NLMS algorithm) , variable step-size update formula can be written W (n+ 1) = w (n)+ e (n) x (n) = w (n)+ (3.1) where, = e (n) x (n) the right to express filter update vector iterative adjust the volume. In order to achieve the purpose of fast convergence, we must choose the appropriate value of variable step size, a possible strategy is as much as possible to reduce the instantaneous squared error, which uses the instantaneous squared error as the mean square error MSE of the simple estimate, which is the basic LMS algorithm思想.
Platform: |
Size: 3072 |
Author: 闫丰 |
Hits:
Description: 在对一些变步长LMS算法分析的基础上,提出了步长因子 (n)与误差信号e(n)之间一种新的非线性函数关系-In a number of variable step size LMS algorithm based on the analysis, put forward a step length factor (n) and error signal e (n) between the non-linear function of a new relationship
Platform: |
Size: 143360 |
Author: 李宁 |
Hits:
Description: 功能描述:测试LMS与RLS算法,比较两种算法的收敛特性
文件名:LMS_RLS_sim.m
测试用例:
x(n)+a1*x(n-1)+a2*x(n-2)=e(n),a1=-1.6,a2=0.81,e(n)为高斯白噪声
文件输出:系数a1的值
调用函数:function [A] = LMS_Algo(M,N,mu,xn)
被调用:无
作者:mingcheng
编写时间:2009-10-13
修改时间:2009-10-13
版本:V1.0 - Function Description: Test LMS and RLS algorithm, the convergence characteristics were compared file name: LMS_RLS_sim.m test case: x (n)+ a1* x (n-1)+ a2* x (n-2) = e (n), a1 =- 1.6, a2 = 0.81, e (n) is Gaussian white noise file output: the value of coefficient a1 call the function: function [A] = LMS_Algo (M, N, mu, xn) is called: No of: mingcheng write time :2009-10-13 modified :2009-10-13 version: V1.0
Platform: |
Size: 1024 |
Author: 赵明诚 |
Hits:
Description: Avetis Ioannisyan
avetis@60ateight.com
Last Updated: 11/30/05
LMS Channel Adaptation
reset randomizers
randn( state ,sum(100*clock))
rand( state ,sum(100*clock))
numPoints = 5000
numTaps = 10 channel order
Mu = 0.001:0.001:0.01 iteration step size
input is guassian
x = randn(numPoints,1) + j*randn(numPoints,1)
choose channel to be random uniform
h = rand(numTaps, 1) + i*rand(numTaps, 1)
h = [1 0 0 0 1] testing only
h = h/max(h) normalize channel
convolve channel with the input
d = filter(h, 1, x)
initialize variables
w = []
y = []
in = []
e = [] error, f- Avetis Ioannisyan
avetis@60ateight.com
Last Updated: 11/30/05
LMS Channel Adaptation
reset randomizers
randn( state ,sum(100*clock))
rand( state ,sum(100*clock))
numPoints = 5000
numTaps = 10 channel order
Mu = 0.001:0.001:0.01 iteration step size
input is guassian
x = randn(numPoints,1) + j*randn(numPoints,1)
choose channel to be random uniform
h = rand(numTaps, 1) + i*rand(numTaps, 1)
h = [1 0 0 0 1] testing only
h = h/max(h) normalize channel
convolve channel with the input
d = filter(h, 1, x)
initialize variables
w = []
y = []
in = []
e = [] error, f
Platform: |
Size: 1024 |
Author: josh |
Hits:
Description: clear all
clc
t=0:1/1000:10-1/1000
s=sin(2*pi*t)
snr=20
s_power=var(s) varience of s
linear_snr=10^(snr/10)
factor=sqrt(s_power/linear_snr)
noise=randn(1,length(s))*factor
x=s+noise Ó É SNR¼ Æ Ë ã Ë æ » úÔ ë É ù
x1=noise Ô ë É ùÔ ´ Ê ä È ë
x2=noise
w1=0 È ¨Ï µ Ê ý ³ õ Ö µ
w2=0
e=zeros(1,length(x))
y=0
u=0.05
for i=1:10000 LMSË ã ·¨
y(i)=w1*x1(i)+w2*x2(i)
e(i)=x(i)-y(i)
w1=w1+u*e(i)*x1(i)
w2=w2+u*e(i)*x2(i)
end
figure(1)
subplot(4,1,1) -clear all
clc
t=0:1/1000:10-1/1000
s=sin(2*pi*t)
snr=20
s_power=var(s) varience of s
linear_snr=10^(snr/10)
factor=sqrt(s_power/linear_snr)
noise=randn(1,length(s))*factor
x=s+noise Ó É SNR¼ Æ Ë ã Ë æ » úÔ ë É ù
x1=noise Ô ë É ùÔ ´ Ê ä È ë
x2=noise
w1=0 È ¨Ï µ Ê ý ³ õ Ö µ
w2=0
e=zeros(1,length(x))
y=0
u=0.05
for i=1:10000 LMSË ã ·¨
y(i)=w1*x1(i)+w2*x2(i)
e(i)=x(i)-y(i)
w1=w1+u*e(i)*x1(i)
w2=w2+u*e(i)*x2(i)
end
figure(1)
subplot(4,1,1)
Platform: |
Size: 1024 |
Author: dasu |
Hits:
Description: 用MATLAB实现LMS自适应滤波器的e(n)^2的曲线及清楚LMS算法-LMS adaptive filter using MATLAB implementation of the e (n) ^ 2 and clear the curve of LMS algorithm
Platform: |
Size: 1024 |
Author: 宁冰 |
Hits:
Description: This derivation of the normalised least mean square algorithm is based on Farhang-
Boroujeny 1999, pp.172-175, and Diniz 1997, pp 150-3. To derive the NLMS algorithm
we consider the standard LMS recursion, for which we select a variable step size
parameter, μ(n). This parameter is selected so that the error value , e+(n), will be
Platform: |
Size: 6144 |
Author: bahtiar |
Hits:
Description: 最小均方算法lms在波束形成中的应用 LMS算法步骤: 1,、设置变量和参量: X(n)为输入向量,或称为训练样本 W(n)为权值向量 b(n)为偏差 d(n)为期望输出 y(n)为实际输出 η为学习速率 n为迭代次数 2、初始化,赋给w(0)各一个较小的随机非零值,令n=0 3、对于一组输入样本x(n)和对应的期望输出d,计算 e(n)=d(n)-X^T(n)W(n) W(n+1)=W(n)+ηX(n)e(n) 4、判断是否满足条件,若满足算法结束,若否n增加1,转入第3步继续执行。-Lms least mean square algorithm applied in Beamforming
Platform: |
Size: 1024 |
Author: 林朝 |
Hits:
Description: 1,、设置变量和参量:
X(n)为输入向量,或称为训练样本
W(n)为权值向量
e(n)为偏差
d(n)为期望输出
y(n)为实际输出
η为学习速率
n为迭代次数
2、初始化,赋给w(0)各一个较小的随机非零值,令n=0
3、对于一组输入样本x(n)和对应的期望输出d,计算
e(n)=d(n)-X^T(n)W(n)
W(n+1)=W(n)+ηX(n)e(n)
4、判断是否满足条件,若满足算法结束,若否n增加1,转入第3步继续执行-, set the variables and parameters:
X (n) is the input vector, otherwise known as the training sample
W (n) for the weight vector
e (n) for the deviation
d (n) is the desired output
y (n) is the actual output
η is the learning rate
n is the number of iterations
Platform: |
Size: 1024 |
Author: 周永辉 |
Hits:
Description: With the advent of large industrial machinery, lighter
weight materials for transportation vehicles and closer
proximity of homes in residential areas, noise
disturbances in listening spaces are becoming even
more prevalent now than in the past. Passive means (i.e.,
physical barriers) to attenuate the noises have been
employed.
Platform: |
Size: 187392 |
Author: ammu |
Hits:
Description: LMS自适应算法
自适应横向滤波器两个权值,输入随机信号r(n)的样本间相互独立,且它的平均功率为Pr =E[r2(n)]=0.01,信号周期为N=16个样点。求最佳权向量解ω0和收敛因子μ的取值范围,并分别汇出ω(0)=[0 0]T,μ=0.1及ω(0)=[4 -10]T,μ=0.05时,两种情况下的权值变化轨迹和第一种情况下误差e(n)与迭代次数n的关系曲线。-LMS adaptive algorithm
The right value of the change trajectory
Platform: |
Size: 3072 |
Author: xinlan |
Hits:
Description: 实现了自适应信号处理LMS算法,分别绘出当w(0)=[0,0]T, =0.1及w(0)=[4, 10]T, =0.05时,LMS算法在两种下的权值变化轨迹、误差e(n)与迭代次数n的关系曲线-The LMS adaptive signal processing algorithm
Platform: |
Size: 1024 |
Author: huangjian |
Hits:
Description: 2017电子竞赛e题软件部分,fpga实现(lms adaptive filter undergraduate electronic design contest)
Platform: |
Size: 10816512 |
Author: 史-诗 |
Hits: