PNN.zip_pnn

%*************************************************************************% % The GAs based Polynominal Neural Networks model % % Encoding: % - Binary % % Genetic operators: % - Roulette-wheel selection % - One-point crossover % - Simple mutation % % Other strategies: % - Elitist strategy % % Remarks: % % Revision 1.2 1999/12/01 % Copyright (c) 2000 by Prof. Gang-Gyoo Jin, Korea Maritime University %*************************************************************************% clear; clc; warning off; delay_u = 3; delay_y = 3; [X,Y]=data_rd(delay_u, delay_y); [X_te Y_te] = data_rd_te(delay_u, delay_y); % initializes the generation gen= 0; [rseed,maxmin,maxgen,popsize,npara,pcross,pmutat,lsubstr,xlb,xub,lchrom,SelNo,... ElitePop,ElitePopInfo,ElitePopY,ElitePopY_te]= bInitPa; pop= bInitPop(rseed,popsize,lchrom); para=UseOfChrom(pop,popsize,npara,lsubstr,xlb,xub,lchrom); [objfunc,Y_data,Y_data_te]= EvalObj(para,popsize,X,Y,X_te,Y_te); fitness= EvalFit(objfunc(:,1),popsize); [chrombest,xbest,objbest,fitbest,fitave,Ybest,Ybest_te]= ... bStatPop(pop,para,objfunc,fitness,Y_data,Y_data_te,maxmin); stats(gen+1,:)=[gen fitbest objbest xbest]; [ElitePop,ElitePopInfo,ElitePopY,ElitePopY_te]=... PopOfElite(pop,popsize,fitness,objfunc,para,Y_data,Y_data_te,SelNo,... ElitePop,ElitePopInfo,ElitePopY,ElitePopY_te,gen); for gen= 1:maxgen fprintf('gen= %d ',gen); if(rem(gen,10) == 0) fprintf('\n'); end if (gen==7) aaaaaa=1; end pop= RwSel(pop,popsize,fitness); [pop,nxover]= bOpXover(pop,popsize,lchrom,pcross); [pop,nmutat]= bSimpMut(pop,popsize,lchrom,pmutat); %********************************************************************** [aa mm]=size(X); sum_pop = sum(pop(:,1:mm),2); for i=1:popsize if(sum_pop(i,1)==0) pop(i,1)=1; end end %********************************************************************** para=UseOfChrom(pop,popsize,npara,lsubstr,xlb,xub,lchrom); [objfunc,Y_data,Y_data_te]= EvalObj(para,popsize,X,Y,X_te,Y_te); fitness= EvalFit(objfunc(:,1),popsize); [pop,para,fitness,objfunc,Y_data,Y_data_te]= ... bElitism(pop,para,fitness,objfunc,Y_data,Y_data_te,... chrombest,xbest,fitbest,objbest,Ybest,Ybest_te,maxmin); [chrombest,xbest,objbest,fitbest,fitave,Ybest,Ybest_te]= ... bStatPop(pop,para,objfunc,fitness,Y_data,Y_data_te,maxmin); stats(gen+1,:)=[gen fitbest objbest xbest]; [ElitePop,ElitePopInfo,ElitePopY,ElitePopY_te]=... PopOfElite(pop,popsize,fitness,objfunc,para,Y_data,Y_data_te,SelNo,... ElitePop,ElitePopInfo,ElitePopY,ElitePopY_te,gen); end subplot(2,1,1) plot(stats(:,1),stats(:,2)) % plots the variables of the best chromosome subplot(2,1,2) plot(stats(:,1),stats(:,3:5))