廣義最小方差自校正控制直接算法

1、%廣義最小方差自校正控制(直接算法)考慮如下系統(tǒng)：y(k)-1.7y(k-1)0.7y(k-2)=u(k-4)0.5u(k-5)(k)0.2(k-1)式中Hk)為方差為0.1的白噪聲。取P(z,)=1,R(z,)=1,Q(z,)=2,期望輸出yr(k)為幅值為10的方波信號(hào)。clearall;closeall;a=1-1.70.7;b=12;c=10.2;d=4;na=length(a)-1;nb=length(b)-1;nc=length(c)-1;nf=nb+d-1;ng=na-1;Pw=1;R=1;Q=2;%加權(quán)多項(xiàng)式np=length(Pw)-1;nr=length(R)-1;nq=l

2、ength(Q)-1;L=400;uk=zeros(d+nf,1);yk=zeros(d+ng,1);yek=zeros(nc,1);yrk=zeros(nc,1);xik=zeros(nc,1);%xiek=zeros(nc,1);yr=10*ones(L/4,1);-ones(L/4,1);ones(L/4,1);-ones(L/4+d,1);xi=sqrt(0.1)*randn(L,1);%RELS初值設(shè)定thetaek=zeros(na+nb+d+nc,d);P=10八6*eye(na+nb+d+nc);fork=1:Ltime(k)=k;y(k)=-a(2:na+1)*yk(1:na

3、)+b*uk(d:d+nb)+c*xi(k);xik;%遞推最小二乘法phie=yk(d:d+ng);uk(d:d+nf);-yek(1:nc);K=P*phie/(1+phie'*P*phie);thetae(:,k)=thetaek(:,1)+K*(y(k)-phie'*thetaek(:,1);P=(eye(na+nb+d+nc)-K*phie')*P;ye=phie'*thetaek(:,d);%最優(yōu)預(yù)測(cè)輸出估計(jì)值%xie=y(k)-phie'*thetae(:,k);%白噪聲估計(jì)值%提取辨識(shí)參數(shù)ge=thetae(1:ng+1,k)'f

4、e=thetae(ng+2:ng+nf+2,k)'ce=1thetae(ng+nf+3:ng+nf+2+nc,k)'ifabs(ce(2)>0.9ce(2)=sign(ce(2)*0.9;endiffe(1)<0.1fe(1)=0.1;end%e,f,g=singlediophantine(ae,be,ce,d);CQ=conv(ce,Q);FP=conv(fe,Pw);CR=conv(ce,R);GP=conv(ge,Pw);u1=-Q(1)*CQ(2:nc+nq+1)*uk(1:nc+nq)/fe(1)-FP(2:np+nf+1)*uk(1:np+nf);u2=

5、CR*yr(k+d:-1:k+d-min(d,nr+nc);yrk(1:nr+nc-d);u(k)=(u1+u2-GP*y(k);yk(1:np+ng)/(Q(1)*Q(1)/fe(1)+fe(1);%更新數(shù)據(jù)fori=d:-1:2thetaek(:,i)=thetaek(:,i-1);endthetaek(:,1)=thetae(:,k);fori=d+nf:-1:2uk(i)=uk(i-1);enduk(1)=u(k);fori=d+ng:-1:2yk(i)=yk(i-1);endyk(1)=y(k);fori=nc:-1:2yek(i)=yek(i-1);yrk(i)=yrk(i-1);

6、endifnc>0yek(1)=ye;yrk(1)=yr(k);xik(1)=xi(k);endendfigure(1);subplot(2,1,1);plot(time,yr(1:L),'r:',time,y);xlabel('k');ylabel('y_r(k),y(k)');legend('y_r(k)','y(k)');axis(0L-2020);subplot(2,1,2);plot(time,u);xlabel('k');ylabel('u(k)');axis(0L

7、-1010);figure(2);subplot(211);plot(1:L,thetae(1:ng+1,:),1:L,thetae(ng+nf+3:ng+2+nf+nc,:);xlabel('k');ylabel('參數(shù)估計(jì)g,c');legend('g_0','g_1','c_1');axis(0L-34);subplot(212);plot(1:L,thetae(ng+2:ng+2+nf,:);xlabel('k');ylabel('參數(shù)估計(jì)f);klegend('f_0','f_1','f_2','f_3','f_4');axis(0L08);106S0n-5-10050100150200250300