系统辨识 递推最小二乘辨识 matlab仿真实验

A. The selected model

The system model：

B. Identification algorithm used

1.Recursive least squares identification. 2.Recursive least square identification with forgetting factor： Take the forgetting factor

clear all
close all
clc
a=[1 -1.3 0.5]';b=[0.8 0.3]';d=3;  %实际系统参数与纯迟延
na=length(a)-1;nb=length(b)-1;  %%na,nb 为输出输入系数矩阵 A,B 的阶数
L=2000;%仿真长度
uk=zeros(d+nb,1);yk=zeros(na,1);  %输入输出初始化
yk_m=zeros(na,1); %模型输出
u=randn(L,1);  %输入信号，采用方差为1的白噪声序列
xi=sqrt(0.1)*randn(L,1);  %干扰信号，采用方差为0.1的白噪声序列

thetae_1=zeros(na+nb+1,1);  %参数初值
P=10^6*eye(na+nb+1);  %修正系数初值
lambda=0.996;  %遗忘因子范围[0.9 1],遗忘因子=1时，即为标准递推最小二乘

for k=1:L
    if k==801
        a=[1 -1.0 0.3]';b=[0.6 0.2]';  %系统参数突变
    end
    theta(:,k)=[a(2:na+1);b];  %系统参数真值
    
    phi=[-yk;uk(d:d+nb)];  %输出输入矩阵
    phi_e=[-yk_m;uk(d:d+nb)]; %模型输出输入矩阵
    y(k)=phi'*theta(:,k)+xi(k);  %系统实际输出
    y_m(k)=phi_e'*thetae_1; %模型输出
    
    %递推公式
    K=P*phi/(lambda+phi'*P*phi);
    thetae(:,k)=thetae_1+K*(y(k)-phi'*thetae_1);
    P=(eye(na+nb+1)-K*phi')*P/lambda;
    
    %更新数据
    thetae_1=thetae(:,k);
    for i=d+nb:-1:2
        uk(i)=uk(i-1);
    end
    uk(1)=u(k);
    
    for i=na:-1:2
        yk(i)=yk(i-1);
        yk_m(i)=yk_m(i-1);
    end
    yk(1)=y(k);
    yk_m(1)=y_m(k);
end
figure('name','参数a')
plot([1:L],thetae(1:na,:));hold on;plot([1:L],theta(1:na,:),'k:');
xlabel('k');ylabel('参数估计a');
legend('a_1','a_2');axis([0 L -2 1]);
figure('name','参数b')
plot([1:L],thetae(na+1:na+nb+1,:));hold on;plot([1:L],theta(na+1:na+nb+1,:),'k:');
xlabel('k');ylabel('参数估计b');
legend('b_0','b_1');axis([0 L 0 1]);
figure('name','输出')
plot([1:L],y,[1:L],y_m,'r');
legend('实际输出','模型输出');
axis([0 L -10 10]);
grid on;

C. The simulation image

Recursive least squares identification：


Recursive least square identification with forgetting factor： Take the forgetting factor



Take the forgetting factor



Take the forgetting factor

D. Make a table to compare the estimation of parameters with the two identification algorithms