【优化算法】混沌粒子群

​function main()clc;clear all;​%swarm definitiond=2;n=50;smin=1*ones(1,d);smax=100*ones(1,d);m1=1;m2=100;wmin=0.2;wmax=1.2;alpha=2;beta=2;max_iter=1000;​cost_func=@(x) Fac(x);​part=struct();part.pos=[];part.cost=Inf;part.pbest=Inf;part.pos_pbest=[];part.vel=[];​swarm=repmat(part,n,1);​gbest=-Inf;pos_gbest=[];%iniitalizing the swarmfor i=1:n  swarm(i).pos=unifrnd(smin,smax,1,d);  swarm(i).cost=cost_func(swarm(i).pos);  swarm(i).vel=rand(1,d);  swarm(i).pos_pbest=swarm(i).pos;  swarm(i).pbest=cost_func(swarm(i).pos_pbest);  if swarm(i).pbest>gbest    gbest=swarm(i).pbest;    pos_gbest=swarm(i).pos_pbest;  endend​for t=1:max_iter    n=size(swarm,1);  w=wmax;    for i=1:n  %calculation of personal bests    if (swarm(i).cost > swarm(i).pbest)      swarm(i).pos_pbest=swarm(i).pos;      swarm(i).pbest=swarm(i).cost;    end  %calculation of global best      if (swarm(i).pbest > gbest)      pos_gbest=swarm(i).pos_pbest;      gbest=swarm(i).pbest;    end     [~, SortOrder]=sort([swarm.cost]);    swarm=swarm(SortOrder);    swarm=flipud(swarm);  %calculation of new inertia weight factor      avg=sum([swarm.cost])/n;    if swarm(i).cost>avg      w=wmin+((wmax-wmin)*(swarm(i).cost-swarm(1).cost)/(avg-swarm(1).cost));    else      w=wmax;    end      %update valocities and positions    swarm(i).vel=w*swarm(i).vel+alpha*rand(1,d).*(pos_gbest-swarm(i).pos)+beta*rand(1,d).*(swarm(i).pos_pbest-swarm(i).pos);      swarm(i).pos=swarm(i).pos+swarm(i).vel;        swarm(i).pos=max(swarm(i).pos,smin);    swarm(i).pos=min(swarm(i).pos,smax);        swarm(i).cost=cost_func(swarm(i).pos);                      end    %retain best n/5 particles  [~, SortOrder]=sort([swarm.cost]);  swarm=swarm(SortOrder);  swarm=flipud(swarm);  swarm2=swarm((1+(1*n/5)):n,:);  swarm=swarm(1:(1*n/5),:);   %chaotic local search for best particle out of n/5 retained particles  temp=swarm(1).pos;% disp ([num2str(cost_func(temp))]);% clsv=(swarm(1).pos-smin)./(smax-smin);%  clsv_n=1*clsv.*(1-clsv);%  if cost_func(smin+clsv_n.*(smax-smin))>swarm(1).cost%  swarm(1).pos=smin+clsv_n.*(smax-smin);%  end% swarm(1).pos=min(swarm(1).pos,smax);%swarm(1).pos=max(swarm(1).pos,smin);  %  swarm(1).cost=cost_func(swarm(1).pos);%  disp ([num2str(swarm(1).cost)]);​  clsv=(pos_gbest-smin)./(smax-smin);  clsv_n=4*clsv.*(1-clsv);%  if cost_func(smin+clsv_n.*(smax-smin))>swarm(1).cost  pos_gbest=smin+clsv_n.*(smax-smin);%  end pos_gbest=min(pos_gbest,smax); pos_gbest=max(pos_gbest,smin); gbest=cost_func(pos_gbest); %  disp ([num2str(smin) '           ' num2str(smax)])  l2=size(swarm2,1);    %new values of min and max  %r=rand(1,d);  r=0.6;  smin=max(smin,pos_gbest-r*(smax-smin));   % smin=min((m1+m2)/2,smin);  smax=min(smax,pos_gbest+r*(smax-smin)); % smax=max((m1+m2)/2,smax);  ​  %get random 4n/5 particles  new_swarm=repmat(part,l2,1);  for i=1:l2    new_swarm(i).pos=unifrnd(smin,smax,1,d);    new_swarm(i).cost=cost_func(new_swarm(i).pos);    new_swarm(i).pbest=swarm2(i).pbest;    new_swarm(i).pos_pbest=swarm2(i).pos_pbest;    new_swarm(i).vel=rand(1,d);  end    swarm=[swarm         new_swarm];   %  w=w*wdamp;  disp(['Iteration : ' num2str(t) ' Minimum value = ' num2str(swarm(1).cost) ' Best solution is : ' num2str(swarm(1).pos)]);endfunction y = Fac(x)​  x1 = x(1);          x2 = x(2);  y = mod((x1*x2),6001);    end

往期回顾>>>>>>

【模式识别】Matlab指纹识别

【优化求解】A*算法解决三维路径规划问题

  matlab自动识别银行卡号

【优化问题】基于栅格地图——遗传算法的机器人最优路径规划

【优化求解】模拟退火遗传实现带时间窗的车辆路径规划问题

【数学建模】Matlab实现SEIR模型

​完整代码加QQ 1575304183