專家PID控制系統(tǒng)Matlab仿真

專家PID控制系統(tǒng)Matlab仿真摘要：分析了一個(gè)速度控制器的控制仿真。其傳遞函數(shù)為：523500G(s)- s3+87.35s2+1047s使用專家PID控制系統(tǒng)，輸入信號(hào)為階躍信號(hào)，取采樣時(shí)間為1ms,畫(huà)出階躍響應(yīng)曲線和誤差變化曲線。0引言專家控制(ExpertControl)的實(shí)質(zhì)是基于受控對(duì)象和控制規(guī)律的各種知識(shí)，并以智能的方式利用這些知識(shí)來(lái)設(shè)計(jì)控制器。利用專家經(jīng)驗(yàn)來(lái)設(shè)計(jì)PID參數(shù)便構(gòu)成專家PID控制。典型的二階系統(tǒng)單位階躍響應(yīng)誤差曲線如圖1、2所示。對(duì)于典型的二階系統(tǒng)階躍響應(yīng)過(guò)程作如下分析，根據(jù)誤差及其變化，可設(shè)計(jì)專家PID控制器，該控制器可分為五種情況進(jìn)行設(shè)計(jì)。1設(shè)計(jì)根據(jù)誤差及其變化，可設(shè)計(jì)專家PID控制器，該控制器可分為以下五種情況進(jìn)行設(shè)計(jì)：(1)當(dāng)1e(k)|〉M1時(shí)，說(shuō)明誤差的絕對(duì)值已經(jīng)很大，不論誤差變化趨勢(shì)如何，都應(yīng)考慮控制器的輸出應(yīng)按最大〔或最小)輸出，以達(dá)到迅速調(diào)整誤差，使誤差絕對(duì)值以最大速度減小。此時(shí)，它相當(dāng)于實(shí)施開(kāi)環(huán)控制。(2)當(dāng)e(k)Ae(k)〉0時(shí)，說(shuō)明誤差在朝誤差絕對(duì)值增大方向變化，或誤差為某一常值，未發(fā)生變化。此時(shí)，如果|e(k)|>M2，說(shuō)明誤差也較大，可考慮由控制器實(shí)施較強(qiáng)的控制作用，以達(dá)到扭轉(zhuǎn)誤差絕對(duì)值朝減小方向變化，并迅速減小誤差的絕對(duì)值。此時(shí)，如果|e(k)|<M2，說(shuō)明盡管誤差朝絕對(duì)值增大方向變化，但誤差絕對(duì)值本身并不很大，可考慮控制器實(shí)施一般的控制作用，只要扭轉(zhuǎn)誤差的變化趨勢(shì)，使其初誤差絕對(duì)值減小方向變化。(3)當(dāng)e(k)Ae(k)<0，e(k)Ae(k-1)>0或e(k)-0時(shí)，說(shuō)明誤差的絕對(duì)值朝減小的方向變化，或者已經(jīng)達(dá)到平衡狀態(tài)。此時(shí)，可考慮采取保持控制器輸出不變。(4)當(dāng)e(k)Ae(k)<0，e(k)Ae(k-1)<0時(shí)，說(shuō)明誤差處于極值狀態(tài)。如果此時(shí)誤差的絕對(duì)值較大，可考慮實(shí)施較強(qiáng)的控制作用。(5)當(dāng)1e(k)|<￡時(shí)，說(shuō)明誤差的絕對(duì)值很小，此時(shí)加入積分，減少穩(wěn)態(tài)誤差。2仿真圖圖1階躍響應(yīng)曲線圖2誤差曲線3程序%ExpertPIDControllerclearall;closeall;ts=0.001;sys=tf(5.235e005,[1,87.35,1.047e004,0]);dsys=c2d(sys,ts,'z');[num,den]=tfdata(dsys,'v');u_1=0.0;u_2=0.0;u_3=0.0;y_1=0;y_2=0;y_3=0;x=[0,0,0]';x2_1=0;kp=0.6;ki=0.03;kd=0.01;error_1=0;fork=1:1:500time(k)=k*ts;rin(k)=1.0; %TracingJieyueSignalu(k)=kp*x(1)+kd*x(2)+ki*x(3);%PIDController%Expertcontrolruleifabs(x(1))>0.8 %Rule1:Unclosedcontrolfirstlyu(k)=0.45;elseifabs(x(1))>0.40u(k)=0.40;elseifabs(x(1))>0.20u(k)=0.12;elseifabs(x(1))>0.01u(k)=0.10;endifx(1)*x(2)>0|(x(2)==0) %Rule2ifabs(x(1))>=0.05u(k)=u_1+2*kp*x(1);elseu(k)=u_1+0.4*kp*x(1);endendif(x(1)*x(2)<0&x(2)*x2_1>0)|(x(1)==0)%Rule3u(k)=u(k);endifx(1)*x(2)<0&x(2)*x2_1<0 %Rule4ifabs(x(1))>=0.05u(k)=u_1+2*kp*error_1;elseu(k)=u_1+0.6*kp*error_1;endendifabs(x(1))<=0.001%Rule5:IntegrationseparationPIcontrolu(k)=0.5*x(1)+0.010*x(3);end%Restrictingtheoutputofcontrollerifu(k)>=10u(k)=10;endifu(k)<=-10u(k)=-10;end%Linearmodelyout(k)=-den(2)*y_1-den(3)*y_2-den(4)*y_3+num⑴*u(k)+num(2)*u_1+num(3)*u2+num(4)*u_3;error(k)=rin(k)-yout(k);% ReturnofPIDparameters %u3=u2;u2=u1;u1=u(k);y_3=y_2;y_2=y_1;y_1=yout(k);x(1)=error(k); %Calculating Px2_1=x(2);x(2)=(error(k)-error_1)/ts; %CalculatingDx(3)=x(3)+error(k)*ts; %CalculatingIerror_1=error(k);endfigure(1);plot(time,rin,'b',time,yout,'r');xlabel('time(s)');ylabel('rin,yout');figure(2);plot(time,rin-yout,'r');xlabel('time(s)');ylabel('error');參考文獻(xiàn)：[1]董海鷹.智能控制理論及應(yīng)用[M].中國(guó)鐵道出版社：北京.2006,47-48(頁(yè))摘要：分析了一個(gè)速度控制器的控制仿真。其傳遞函數(shù)為：523500G(s)- s3+87.35s2+1047s使用模糊PID控制系統(tǒng)，輸入信號(hào)為階躍信號(hào)，取采樣時(shí)間為1ms,畫(huà)出階躍響應(yīng)曲線和誤差變化曲線。0引言模糊PID控制算法以誤差E和Ec為輸入，利用模糊控制規(guī)則在線對(duì)PID參數(shù)進(jìn)行修改，以滿足不同時(shí)刻的E和Ec對(duì)PID參數(shù)的要求，輸出為Kp，Ki，Kd。圖1模糊控制原理圖該模糊推理控制系統(tǒng)是一個(gè)兩輸入三輸出的形式，輸入變量為E和Ec，輸出變量為Kp，Ki和Kd，其調(diào)整原則為：⑴當(dāng)誤差|E|較大時(shí)，為使系統(tǒng)具有較好的快速跟蹤性能，不論誤差的變化趨勢(shì)如何，都應(yīng)該取較大的Kp和較小的Kd，同時(shí)為避免系統(tǒng)響應(yīng)出現(xiàn)較大的超調(diào)，應(yīng)對(duì)積分作用加以限制，取較小的Ki值。⑵當(dāng)誤差|E|處于中等大小時(shí)，為使系統(tǒng)響應(yīng)具有較小的超調(diào)，Kp應(yīng)取得小些,同時(shí)為保證系統(tǒng)的響應(yīng)速度，Ki和Kd的大小要適中，其中Kd的取值對(duì)系統(tǒng)響應(yīng)的影響較大。⑶當(dāng)誤差|E|較小時(shí)，為保證系統(tǒng)具有較好的穩(wěn)態(tài)性能，Kp和Ki應(yīng)取得大些，同時(shí)為避免系統(tǒng)在設(shè)定值附近出現(xiàn)振蕩，并考慮系統(tǒng)的抗干擾性能，當(dāng)|E|較小時(shí)，Kd可取大些；當(dāng)|E|較大時(shí)，Kd應(yīng)取小些。1建立控制規(guī)則表輸入變量E，Ec和Kd語(yǔ)言值的模糊子集取為｛負(fù)大，負(fù)中，負(fù)小，零，正小,正中，正大｝，并簡(jiǎn)記為｛NB，NM，NS，ZO，PS，PM，PB｝，論域?yàn)椋?3，-2，-1，0，1，2，3｝。Ki論域?yàn)椋?0.06，-0.04,-0.02，0，0.02，0.04，0.06｝，Kp論域?yàn)椋?0.3,-0.2,-0.l，0,0.1,0.2,0.3｝。隸屬函數(shù)采用三角形函數(shù),根據(jù)對(duì)已有控制系統(tǒng)設(shè)計(jì)經(jīng)驗(yàn)的總結(jié)，得到PID三個(gè)控制參數(shù)的模糊控制規(guī)則表，見(jiàn)表1?表3。eAkpNBNM表1Kp的模糊規(guī)則表PSPMPBNSZOdeNBPBPBPMPMPSZOZONMPBPBPMPSPSZONSNSPMPMPMPSZONSNS-ZO PM PM——-1 PS zo L——NS NM NMPSPSPSZONSNSNMNMPMPSZONSNMNMNMNBnnNENEPBZOZONMNMNMNBNB注：NB、NM、NS、ZO、PS、PM、PB分別表示負(fù)大、負(fù)中、負(fù)小、零、正小、正中、正大。表2Ki的模糊規(guī)則表e△kiNBNMNSZOPSPMPBdeNBNBNBNMNMNSZOZONMNBNBNMNSNSZOZONSNBNMNSNSZOPSPSZONMNMNSZOPSPMPMPSNMNSZOPSPSPMPBPMZOZOPSPSPMPBPBpnphphPBZOZOPSPMPMPBPB表3Kd的模糊規(guī)則表e△kdNBNMNSZOPSPMPBdeNBPSNSNBNBNBNMPSNMPSNSNBNMNMNSZONSZONSNMNMNSNSZOZOZONSNSNSNSNSZOPSZOZOZOZOZOZOZOPMPBNSPSPSPSPSPB PB PB PM PM PM PS PS PB 2仿真結(jié)果

圖3kp的隸屬度函數(shù)曲線

圖5e的隸屬度函數(shù)曲線

圖7ki的曲線

圖8u的輸出曲線圖10階躍響應(yīng)的曲線Systemfuzzpid:2inputs,3outputs,49rules圖11規(guī)則輸入輸出3輸出49條規(guī)則ans=If(eisNB)and(ecisNB)then(kpisPB)(kiisNB)(kdisPS)⑴If(eisNB)and(ecisNM)then(kpisPB)(kiisNB)(kdisNS)(1)3.If(eisNB)and(ecisNS)then(kpisPM)(kiisNM)(kdisNB)(1)4.If(eisNB)and(ecisZ)then(kpisPM)(kiisNM)(kdisNB)(1)5.If(eisNB)and(ecisPS)then(kpisPS)(kiisNS)(kdisNB)(1)6.If(eisNB)and(ecisPM)then(kpisZ)(kiisZ)(kdisNM)(1)7.If(eisNB)and(ecisPB)then(kpisZ)(kiisZ)(kdisPS)(1)If(eisNM)and(ecisNB)then(kpisPB)(kiisNB)(kdisPS)(1)If(eisNM)and(ecisNM)then(kpisPB)(kiisNB)(kdisNS)(1)10.If(eisNM)and(ecisNS)then(kpisPM)(kiisNM)(kdisNB)(1)11.If(eisNM)and(ecisZ)then(kpisPS)(kiisNS)(kdisNM)(1)12.If(eisNM)and(ecisPS)then(kpisPS)(kiisNS)(kdisNM)(1)13.If(eisNM)and(ecisPM)then(kpisZ)(kiisZ)(kdisNS)(1)14.If(eisNM)and(ecisPB)then(kpisNS)(kiisZ)(kdisZ)(1)15.If(eisNS)and(ecisNB)then(kpisPM)(kiisNB)(kdisZ)(1)16.If(eisNS)and(ecisNM)then(kpisPM)(kiisNM)(kdisNS)(1)17.If(eisNS)and(ecisNS)then(kpisPM)(kiisNS)(kdisNM)(1)18.If(eisNS)and(ecisZ)then(kpisPS)(kiisNS)(kdisNM)(1)19.If(eisNS)and(ecisPS)then(kpisZ)(kiisZ)(kdisNS)(1)If(eisNS)and(ecisPM)then(kpisNS)(kiisPS)(kdisNS)(1)21.If(eisNS)and(ecisPB)then(kpisNS)(kiisPS)(kdisZ)(1)22.If(eisZ)and(ecisNB)then(kpisPM)(kiisNM)(kdisZ)(1)23.If(eisZ)and(ecisNM)then(kpisPM)(kiisNM)(kdisNS)(1)24.If(eisZ)and(ecisNS)then(kpisPS)(kiisNS)(kdisNS)(1)25.If(eisZ)and(ecisZ)then(kpisZ)(kiisZ)(kdisNS)(1)If(eisZ)and(ecisPS)then(kpisNS)(kiisPS)(kdisNS)(1)If(eisZ)and(ecisPM)then(kpisNM)(kiisPM)(kdisNS)(1)28.If(eisZ)and(ecisPB)then(kpisNM)(kiisPM)(kdisZ)(1)29.If(eisPS)and(ecisNB)then(kpisPS)(kiisNM)(kdisZ)(1)30.If(eisPS)and(ecisNM)then(kpisPS)(kiisNS)(kdisZ)(1)31.If(eisPS)and(ecisNS)then(kpisZ)(kiisZ)(kdisZ)(1)If(eisPS)and(ecisZ)then(kpisNS)(kiisPS)(kdisZ)(1)If(eisPS)and(ecisPS)then(kpisNS)(kiisPS)(kdisZ)(1)If(eisPS)and(ecisPM)then(kpisNM)(kiisPM)(kdisZ)(1)35.If(eisPS)and(ecisPB)then(kpisNM)(kiisPB)(kdisZ)(1)36.If(eisPM)and(ecisNB)then(kpisPS)(kiisZ)(kdisPB)(1)37.If(eisPM)and(ecisNM)then(kpisZ)(kiisZ)(kdisPS)(1)38.If(eisPM)and(ecisNS)then(kpisNS)(kiisPS)(kdisPS)(1)39.If(eisPM)and(ecisZ)then(kpisNM)(kiisPS)(kdisPS)(1)40.If(eisPM)and(ecisPS)then(kpisNM)(kiisPM)(kdisPS)(1)41.If(eisPM)and(ecisPM)then(kpisNM)(kiisPB)(kdisPS)(1)42.If(eisPM)and(ecisPB)then(kpisNB)(kiisPB)(kdisPB)(1)43.If(eisPB)and(ecisNB)then(kpisZ)(kiisZ)(kdisPB)(1)44.If(eisPB)and(ecisNM)then(kpisZ)(kiisZ)(kdisPM)(1)45.If(eisPB)and(ecisNS)then(kpisNM)(kiisPS)(kdisPM)(1)If(eis PB)and (ec is Z)then(kpisNM)(kiisPM)(kdisPM)(1)If(eis PB)and (ec is PS)then(kpisNM)(kiisPM)(kdisPS)(1)If(eis PB)and (ec is PM)then(kpisNB)(kiisPB)(kdisPS)(1)If(eis PB)and (ec is PB)then(kpisNB)(kiisPB)(kdisPB)(1)4程序：%FuzzyTunningPIDControlclearall;closeall;a=newfis('fuzzpid');a=addvar(a,'input','e',[-3,3]);%Parameterea=addmf(a,'input',1,'NB','zmf,[-3,-1]);a=addmf(a,'input',1,'NM','trimf,[-3,-2,0]);a=addmf(a,'input',1,'NS','trimf,[-3,-1,1]);a=addmf(a,'input',1,'Z','trimf',[-2,0,2]);a=addmf(a,'input',1,'PS','trimf,[-1,1,3]);a=addmf(a,'input',1,'PM','trimf,[0,2,3]);a=addmf(a,'input',1,'PB','smf,[1,3]);a=addvar(a,'input','ec',[-3,3]);%Parametereca=addmf(a,'input',2,'NB','zmf,[-3,-1]);a=addmf(a,'input',2,'NM','trimf,[-3,-2,0]);a=addmf(a,'input',2,'NS','trimf,[-3,-1,1]);a=addmf(a,'input',2,'Z','trimf,[-2,0,2]);a=addmf(a,'input',2,'PS','trimf,[-1,1,3]);a=addmf(a,'input',2,'PM','trimf,[0,2,3]);a=addmf(a,'input',2,'PB','smf,[1,3]);a=addvar(a,'output','kp',[-0.3,0.3]);%Parameterkpa=addmf(a,'output',1,'NB','zmf,[-0.3,-0.1]);a=addmf(a,'output',1,'NM','trimf,[-0.3,-0.2,0]);a=addmf(a,'output',1,'NS','trimf,[-0.3,-0.1,0.1]);a=addmf(a,'output',1,'Z','trimf,[-0.2,0,0.2]);a=addmf(a,'output',1,'PS','trimf,[-0.1,0.1,0.3]);a=addmf(a,'output',1,'PM','trimf,[0,0.2,0.3]);a=addmf(a,'output',1,'PB','smf,[0.1,0.3]);a=addvar(a,'output','ki',[-0.06,0.06]);%Parameterkia=addmf(a,'output',2,'NB','zmf,[-0.06,-0.02]);a=addmf(a,'output',2,'NM','trimf,[-0.06,-0.04,0]);a=addmf(a,'output',2,'NS','trimf,[-0.06,-0.02,0.02]);a=addmf(a,'output',2,'Z','trimf,[-0.04,0,0.04]);a=addmf(a,'output',2,'PS','trimf,[-0.02,0.02,0.06]);a=addmf(a,'output',2,'PM','trimf,[0,0.04,0.06]);a=addmf(a,'output',2,'PB','smf,[0.02,0.06]);a=addvar(a,'output','kd',[-3,3]);%Parameterkpa=addmf(a,'output',3,'NB','zmf,[-3,-1]);a=addmf(a,'output',3,'NM','trimf,[-3,-2,0]);a=addmf(a,'output',3,'NS','trimf,[-3,-1,1]);a=addmf(a,'output',3,'Z','trimf,[-2,0,2]);a=addmf(a,'output',3,'PS','trimf,[-1,1,3]);a=addmf(a,'output',3,'PM','trimf,[0,2,3]);a=addmf(a,'output',3,'PB','smf,[1,3]);rulelist=[1171511;1271311;1362111;1462111;1553111;1644211;1744511;2171511;2271311;2362111;2453211;2553211;2644311;2734411;6526511;6627511;6717711;7144711;7244611;7325611;7426611;7526511;7617511;7717711];a=addrule(a,rulelist);a=setfis(a,'DefuzzMethod','centroid');writefis(a,'fuzzpid');3161411;3262311;3363211;3453211;3544311;3635311;3735411;4162411;4262311;4353311;4444311;4535311;4626311;4726411;5152411;5253411;5344411;5435411;5535411;5626411;5727411;6154711;6244511;6335511;6425511;a=readfis('fuzzpid');%PIDControllerts=0.001;sys=tf(5.235e005,[1,87.35,1.047e004,0]);dsys=c2d(sys,ts,'tustin');[num,den]=tfdata(dsys,'v');u_1=0.0;u_2=0.0;u_3=0.0;y_1=0;y_2=0;y_3=0;x=[0,0,0]';error_1=0;e_1=0.0;ec_1=0.0;kp0=0.40;kd0=1.0;ki0=0.0;fork=1:1:500time(k)=k*ts;rin(k)=1;%UsingfuzzyinferencetotunningPIDk_pid=evalfis([e_1,ec_1],a);kp(k)=kp0+k_pid(1);ki(k)=ki0+k_pid(2);kd(k)=kd0+k_pid(3);u(k)=kp(k)*x(1)+kd(k)*x(2)+ki(k)*x(3);ifk==300 %Addingdisturbance(1.0vattime0.3s)u(k)=u(k)+1.0;endifu(k)>=10u(k)=10;endifu(k)<=-10u(k)=-10;endyout(k)=-den(2)*y_1-den(3)*y_2-den(4)*y_3+num⑴*u(k)+num(2)*u_1+num(3)*u_2+num(4)*u_3;error(k)=rin(k)-yout(k);%%%%%%%%%%%%%%ReturnofPIDparameters%%%%%%%%%%%%%%%u_3=u_2;u_2=u_1;u_1=u(k);y_3=y_2;y_2=y_1;y_1=yout(k);CalculatingIe_1=x(1);ec_1=x(2);error_2=error_1;error_1=error(k);endshowrule(a)figure(1);plot(time,rin,'b',time,yout,'r');xlabel('time(s)');ylabel('rin,yout');figure(2);plot(time,error,'r');xlabel('time(s)');ylabel('error');figure(3);plot(time,u,'r');xlabel('time(s)');ylabel('u');figure(4);plot(time,kp,'r');xlabel('time(s)');ylabel('kp');figure(5);plot(time,ki,'r');xlabel('time(s)');ylabel('ki');figure(6);plot(time,kd,'r');xlabel('time(s)');ylabel('kd');figure(7);plotmf(a,'input',1);figure(8);plotmf(a,'input',2);figure(9);plotmf(a,'output',1);figure(10);plotmf(a,'output',2);figure(11);plotmf(a,'output',3);plotfis(a);fuzzyfuzzpidx(1)=error(k); %CalculatingPx(2)=error(k)-error_1; %CalculatingDx(3)=x(3)+error(k); %神經(jīng)網(wǎng)絡(luò)控制系統(tǒng)Matlab仿真摘要：分析了一個(gè)速度控制器的控制仿真。被控制對(duì)象的近似數(shù)學(xué)模型為:yout（k）=+u（k一1）a（k）yout（k一yout（k）=+u（k一1）式中，系數(shù)a(k)是慢時(shí)變的，a(k)=1.2(1-0.8e-0.1k)。0引言神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu)選擇4-5-3,學(xué)習(xí)速率”二0.28和慣性系數(shù)a=0.04，加權(quán)系數(shù)初始值取區(qū)間［-0.5,0.5］上的隨機(jī)數(shù)。輸入指令信號(hào)分為兩種：rin(k)=sin(2;t)。取S=2時(shí)為正弦跟蹤，初始權(quán)值取隨機(jī)值，運(yùn)行穩(wěn)定后用穩(wěn)定權(quán)值代替隨機(jī)值。其跟蹤結(jié)果和相應(yīng)的曲線如圖所示。

圖1正弦跟蹤響應(yīng)曲線（s=12）1分析PID控制要取得較好的控制效果，就必須通過(guò)調(diào)整好比例、積分和微分三種控制作用，形成控制量中既相互配合又相互制約的關(guān)系，這種關(guān)系不一定是簡(jiǎn)單的“線性組合”，從變化無(wú)窮的非線性組合中可以找出最佳的。神經(jīng)網(wǎng)絡(luò)所具有的任意非線性表達(dá)能力，可以通過(guò)對(duì)系統(tǒng)性能的學(xué)習(xí)來(lái)實(shí)現(xiàn)具有最佳組合的PID控制。采用BP網(wǎng)絡(luò)，可以建立參數(shù)ki，kp，kd自學(xué)習(xí)的PID控制器。

圖2誤差曲線圖3輸出響應(yīng)曲線0.40012 3 4 5time(s)6圖4參數(shù)自適應(yīng)整定曲線程序：程序：%BPbasedPIDControlclearall;closeall;xite=0.20;alfa=0.05;S=1;%Signaltype%wi=0.50*rands(H,IN);wi_1=wi;wi_2=wi;wi_3=wi;wo=[0.75760.26160.5820-0.1416-0.1325;-0.11460.29490.83520.22050.4508;0.72010.45660.76720.49620.3632];%wo=0.50*rands(Out,H);wo_1=wo;wo_2=wo;wo_3=wo;endIN=4;H=5;Out=3;%NNStructureifS==1%StepSignalwi=[-0.6394-0.2696-0.3756-0.7023;-0.8603-0.2013-0.5024-0.2596;-1.07490.5543-1.6820-0.5437;-0.3625-0.0724-0.6463-0.2859;0.14250.0279-0.5406-0.7660];ifS==2%SineSignalwi=[-0.28460.2193-0.5097-1.0668;-0.7484-0.1210-0.47080.0988;-0.71760.8297-1.60000.2049;-0.08580.1925-0.63460.0347;0.43580.2369-0.4564-0.1324];

%wi=0.50*rands(H,IN);wi_1=wi;wi_2=wi;wi_3=wi;wo=[1.0438 0.5478 0.86820.1446 0.1537;0.1716 0.5811 1.12140.5067 0.7370;1.0063 0.7428 1.05340.7824 0.6494];%wo=0.50*rands(Out,H);wo_1=wo;wo_2=wo;wo_3=wo;endx=[0,0,0];du_1=0;u_1=0;u_2=0;u_3=0;u_4=0;u_5=0;y_1=0;y_2=0;y_3=0;Oh=zeros(H,1);middlelayerI=Oh;middlelayererror_2=0;error_1=0;ts=0.001;fork=1:1:6000time(k)=k*ts;ifS==1rin(k)=1.0;elseifS==2rin(k)=sin(1*2*pi*k*ts);end%Unlinearmodela(k)=1.2*(1-0.8*exp(-0.1*k));yout(k)=a(k)*y_1/(1+y_1A2)+u_1;error(k)=rin(k)-yout(k);xi=[rin(k),yout(k),