ACO.zip_ant colony

clc clear all close all format shortG %% parameters setting d=load('data.mat'); w=d.w; W=d.W; W=45; v=d.v; nvar=length(w); % number of variable lb=zeros(1,nvar); %ub=[10*ones(1,10) 5*ones(1,10) ones(1,10)]; ub=ones(1,nvar); NL=ub-lb+1; % number of level n_ant=100; % number of ant maxiter=400; % max of iteration %% initialization tau0=1; for j=1:nvar tau{j}=tau0*ones(1,NL(j)); % phromone matrix etha{j}=[w(j)/v(j) v(j)/w(j)]; end % etha=1./dis; % hueristic information matrix % etha=etha-diag(diag(etha)); Q=0.1; alpha=1; beta=1; Roh=0.02; %% main loop tic empty.pos=[]; empty.cost=[]; ant=repmat(empty,n_ant,1); % population gant.pos=[]; % global ant gant.cost=-inf; best=zeros(maxiter,1); AVR=zeros(maxiter,1); for iter=1:maxiter for i=1:n_ant for j=1:nvar P=(tau{j}.^alpha);%(etha(tour(end),:).^beta); k=RouletteWheel(P); ant(i).level(j)=k; ant(i).pos(j)=lb(j)+k-1; end ant(i).cost=fitness(ant(i).pos,nvar,w,W,v); end [value,index]=max([ant.cost]); if value>gant.cost gant=ant(index); end for i=1:n_ant L=ant(i).cost; for j=1:nvar tau{j}(ant(i).level(j))=tau{j}(ant(i).level(j))+(Q*L); end end for j=1:nvar tau{j}=tau{j}*(1-Roh); end best(iter)=gant.cost; AVR(iter)=mean([ant.cost]); disp([ ' iter = ' num2str(iter) ' BEST = ' num2str(best(iter)) ]) if iter>150 && best(iter)==best(iter-150) break end end %% results disp('======================================================') disp([ ' best solution = ' num2str(gant.pos)]); disp([ ' best fitness = ' num2str(gant.cost)]); disp([ ' Time = ' num2str(toc)]) figure(1) plot(best(1:iter),'r','LineWidth',2); hold on plot(AVR(1:iter),'b','LineWidth',2); xlabel(' iteration ') ylabel('fitness') legend('BEST','MEAN') title(' ACO for TSP ')