matlab-基于模糊神经网络PID控制器的matlab仿真-源码

%神经模糊算法的法仿真程序 clc; clear; close all; warning off; addpath(genpath(pwd)); p1=0.5; q1=0.5; p2=0.5; q2=0.5; p3=0.5; q3=0.5; p4=0.95; q4=0.5; p5=0.95; q5=0.5; p6=0.95; q6=0.5; p7=0.5; q7=0.5; p8=0.95; q8=0.5; p9=0.5; q9=0.5; p10=0.95; q10=0.5; p11=0.95; q11=0.5; p12=0.95; q12=0.5; p13=0.65; q13=0.05; %权向量初始化部分 a10=[-2 0 2]; a11=[-2 0 2]; a20=[-2 0 2]; a21=[-2 0 2]; b10=[1.5 1.5 1.5]; b11=[1.5 1.5 1.5]; b20=[1.5 1.5 1.5]; b21=[1.5 1.5 1.5]; v0=[-1 -0.5 -0.5;-0.5 0 0.5;0.5 0.5 1]; v1=[-1 -0.5 -0.5;-0.5 0 0.5;0.5 0.5 1]; %系统部分的初值 yp0=0; yp1=0; ep0=0; ep1=0; up0=1.05; up1=1.39; x1=0.12; x2=0.24; y0=0; y1=0; u1=0; e0=0; e1=0; sp=10; k=1; se=0.02; sd=0.02; su=0.0522; %开始 kp=0.284; ki=0.03; kd=0.626; %the second channel reference p12=0.5; q12=0.5; p22=0.5; q22=0.5; p32=0.5; q32=0.5; p42=0.95; q42=0.5; p52=0.95; q52=0.5; p62=0.95; q62=0.5; p72=0.5; q72=0.5; p82=0.95; q82=0.5; p92=0.5; q92=0.5; p102=0.95; q102=0.5; p112=0.95; q112=0.5; p122=0.95; q122=0.5; p132=0.65; q132=0.05; %系统部分的初值 yp02=0; yp12=0; ep02=0; ep12=0; up02=1.05; up12=1.39; x12=0.12; x22=0.24; y02=0; y12=0; u12=0; e02=0; e12=0; sp2=5; k=1; se2=0.02; sd2=0.02; su2=0.0522; kp2=0.284; ki2=0.03; kd2=0.626; %++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++%子系统1(SUB1)的J循环开始 for J=1:50 ep1=10-yp1; pid=kp*(ep1-ep0)+ki*ep1; up2=up1+pid; yp2=0.5*yp1+2.5*up2+2.5*up1; %yp2=0.8333*yp1-0.1667*yp0+1.6667*up2+1.6667*up1; yp(:,J)=yp2; up0=up1; up1=up2; ep0=ep1; yp0=yp1; yp1=yp2; end time=[1:1:50]; n=0.28*rand(size(time)); %产生的SUB1随机噪声 n1=0.3*rand(size(time));%产生的SUB2随机噪声 while(k<=50) %function deal with for(T=1:20) X1=x1*se; X2=x2*sd; if(X1<=-2) X1=-2; elseif(X1>=2) X1=2; end if(X2<=-2) X2=-2; elseif(X2>=2) X2=2; end %the hiddening layer output of FNN for i=1:3 for j=1:3 A=[(X1-a11(:,i))/b11(:,i)].^2; B=[(X2-a21(:,j))/b21(:,j)].^2; h(i,j)=exp(-(A+B)/2); end end %the output sum=0; for i=1:3 for j=1:3 sum=sum+h(i,j)*v1(i,j); end end ot=sum; cu=su*ot; u2=u1+cu; disp(u2); if(u2<0) u2=0; elseif(u2>=1) u2=1; end y2=0.5*y1+2.5*u2+2.5*u1+n1(:,k)+0.01*y12; %+n1(:,k)+0.01*y12表示随机噪声和子系统间的相互耦合 disp(['the output y number is' int2str(T)]); disp(y2); Y(:,k)=y2; E=0.5*(sp-y2).^2; e2=sp-y2; Es(:,k)=e2; x1=e2; x2=e2-e1; e0=e1; e1=e2; delot=(sp-y2)*2.5*su; for i=1:3 for j=1:3 dv=v1(i,j)-v0(i,j); v2=v1(i,j)+p13*delot+q13*dv; v3(i,j)=v2; end end %refreshing the center and width %a11 s1=0; for j=1:3 s1=s1+v1(1,j)*h(1,j); end pa11=s1; a110=a10(:,1); a111=a11(:,1); dela11=pa11*(X1-a111)/b11(:,1).^2; da11=a111-a110; a112=a111+p1*dela11+q1*da11; a10(:,1)=a111; a11(:,1)=a112; %a12 s2=0; for j=1:3 s2=s2+v1(2,j)*h(2,j); end pa12=s2; a120=a10(:,2); a121=a11(:,2); dela12=pa12*(X1-a121)/b11(:,2).^2; da12=a121-a120; a122=a121+p2*dela12+q2*da12; a10(:,2)=a121; a11(:,2)=a122; %a13 s3=0; for j=1:3 s3=s3+v1(3,j)*h(3,j); end pa13=s3; a130=a10(:,3); a131=a11(:,3); dela13=pa13*(X1-a131)/b11(:,3).^2; da13=a131-a130; a132=a131+p3*dela13+q3*da13; a10(:,3)=a131; a11(:,3)=a132; % b11 pb11=pa11; b110=b10(:,1); b111=b11(:,1); delb11=pb11*[(X1-a111)].^2/b111.^3; db11=b111-b110; b112=b111+p4*delb11+q4*db11; b10(:,1)=b111; b11(:,1)=b112; %b12 pb12=pa12; b120=b10(:,2); b121=b11(:,2); delb12=pb12*[(X1-a121)].^2/b121.^3; db12=b121-b120; b122=b121+p5*delb12+q5*db12; b10(:,2)=b121; b11(:,2)=b122; %b13 pb13=pa13; b130=b10(:,3); b131=b11(:,3); delb13=pb13*[(X1-a131)].^2/b131.^3; db13=b131-b130; b132=b131+p6*delb13+q6*db13; b10(:,3)=b131; b11(:,3)=b132; %a21 s4=0; for i=1:3 s4=s4+v1(i,1)*h(i,1); end pa21=s4; a210=a20(:,1); a211=a21(:,1); dela21=pa21*(X2-a211)/b21(:,1).^2; da21=a211-a210; a212=a211+p7*dela21+q7*da21; a20(:,1)=a211; a21(:,1)=a212; %a22 s5=0; for i=1:3 s5=s5+v1(i,2)*h(i,2); end pa22=s5; a220=a20(:,2); a221=a21(:,2); dela22=pa22*(X2-a221)/b21(:,2).^2; da22=a221-a220; a222=a221+p8*dela22+q8*da22; a20(:,2)=a221; a21(:,2)=a222; %a23 s6=0; for i=1:3 s6=s6+v1(i,3)*h(i,3); end pa23=s6; a230=a20(:,3); a231=a21(:,3); dela23=pa23*(X2-a231)/b21(:,3).^2; da23=a231-a230; a232=a231+p9*dela23+q3*da23; a20(:,3)=a231; a21(:,3)=a232; %b21 pb21=pa21; b210=b20(:,1); b211=b21(:,1); delb21=pb21*[(X2-a211)].^2/b211.^3; db21=b211-b210; b212=b211+p10*delb21+q10*db21; b20(:,1)=b211; b21(:,1)=b212; %b22 pb22=pa22; b220=b20(:,2); b221=b21(:,2); delb22=pb22*[(X2-a221)].^2/b221.^3; db22=b221-b220; b222=b221+p11*delb22+q11*db22; b20(:,2)=b221; b21(:,2)=b222; %b23 pb23=pa23; b230=b20(:,3); b231=b21(:,3); delb23=pb23*[(X2-a231)].^2/b231.^3; db23=b231-b230; b232=b231+p12*delb23+q12*db23; b20(:,3)=b231; b21(:,3)=b232; v0=v1; v1=v3; if(abs(e1)<0.00015) break; end end a11 a21 b11 b21 v3 y0=y1; y1=y2 u0=u1; u1=u2; %adding the second channel %权向量初始化部分 a102=[-2 0 2]; a112=[-2 0 2]; a202=[-2 0 2]; a212=[-2 0 2]; b102=[1.5 1.5 1.5]; b112=[1.5 1.5 1.5]; b202=[1.5 1.5 1.5]; b212=[1.5 1.5 1.5]; v02=[-1 -0.5 -0.5;-0.5 0 0.5;0.5 0.5 1]; v12=[-1 -0.5 -0.5;-0.5 0 0.5;0.5 0.5 1]; %++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++%子系统2(SUB2)的J循环开始 for J=1:100 ep2=5-yp12; pid2=kp2*(ep12-ep02)+ki2*ep12; up22=up12+pid2; yp22=0.5*yp12+1.25*up22+1.25*up12; %yp2=0.8333*yp1-0.1667*yp0+1.6667*up2+1.6667*up1; yp2(:,J)=yp22; up02=up12; up12=up22; ep02=ep12; yp02=yp12; yp12=yp22; end %while(k<=250) %function deal width for(T=1:20) X12=x12*se2; X22=x22*sd2; if(X12<=-2) X12=-2; elseif(X12>=2) X12=2; end if(X22<=-2) X22=-2; elseif(X22>=2) X22=2; end %the hiddening layer output of FNN, for i=1:3 for j=1:3 A=[(X12-a112(:,i))/b112(:,i)].^2; B=[(X22-a212(:,j))/b212(:,j)].^2; h(i,j)=exp(-(A+B)/2); end end %the output sum2=0; for i=1:3 for j=1:3 sum2=sum2+h(i,j)*v12(i,j); end end ot=sum2; cu=su2*ot; u22=u12+cu; disp(u22); if(u22<0) u22=0; elseif(u22>=1) u22=1; end y22=0.5*y12+1.25*u22+1.25*u12+n(:,k)+0.01*y1; %+n(:,k)+0.01*y1表示随机噪声和子系统间的相互耦合 disp(['the output y2 number is' int2str(T)]); disp(y22); Y2(:,k)=y22; E2=0.5*(sp2-y22).^2; e22=sp2-y22; E22(:,k)=e22; x12=e22; x22=e22-e12; e02=e12; e12=e22; delot=(sp2-y22)*1.25*su2; for i=1:3 for j=1:3 dv=v12(i,j)-v02(i,j); v22=v12(i,j)+p132*delot+q132*dv; v32(i,j)=v22; end end %refreshing the center and width %a11 s12=0; for j=1:3 s12=s12+v12(1,j)*h(1,j); end pa112=s12; a1102=a102(:,1); a1112=a112(:,1); dela112=pa112*(X12-a1112)/b112(:,1).^2; da112=a1112-a1102; a1122=a1112+p12*dela112+q12*da112; a102(:,1)=a1112; a112(:,1)=a1122; %a12 s22=0; for j=1:3 s22=s22+v12(2,j)*h(2,j); end pa122=s22; a1202=a102(:,2); a1212=a112(:,2); dela122=pa122*(X12-a121)/b112(:,2).^2; da122=a1212-a1202; a1222=a1212+p22*dela122+q22*da122; a102(:,2)=a1212; a112(:,2)=a1222; %a13 s32=0; for j=1:3 s32=s32+v12(3,j)*h(3,j); end