遺傳算法模擬退火matlab編程

1、單鉆頭退火算法單鉆頭退火算法matlab編程編程clearclca=0.999%溫度衰減函數(shù)的參數(shù)t0=97tf=3t=t0Markov_length=2800%Markov鏈長度codinates=[]codinates(:1)=[]amount=size(codinates1)%城市的數(shù)目%通過向量化的方法計算距離矩陣dist_matrix=zeros(amountamount)co_x_tmp1=codinates(:1)ones

2、(1amount)co_x_tmp2=co_x_tmp1co_y_tmp1=codinates(:2)ones(1amount)co_y_tmp2=co_y_tmp1dist_matrix=sqrt((co_x_tmp1co_x_tmp2).^2...(co_y_tmp1co_y_tmp2).^2)sol_new=1:amount%產(chǎn)生初始解%sol_new是每次產(chǎn)生的新解；sol_current是當前解；sol_best是冷卻中的最好

3、解；E_current=infE_best=inf%E_current是當前解對應的回路距離；%E_new是新解的回路距離；%E_best是最優(yōu)解的sol_current=sol_newsol_best=sol_newp=1whilet=tffr=1:Markov_length%Markov鏈長度%產(chǎn)生隨機擾動if(r0.5)%隨機決定是進行兩交換還是三交換%兩交換ind1=0ind2=0while(ind1==ind2)ind1=ce

4、il(r.amount)ind2=ceil(r.amount)endtmp1=sol_new(ind1)%dist_matrix(sol_new(amount)sol_new(1))ifE_newE_currentE_current=E_newsol_current=sol_newifE_newE_best%把冷卻過程中最好的解保存下來E_best=E_newsol_best=sol_newendelse%若新解的目標函數(shù)值小于當前解的

5、，%則僅以一定概率接受新解ifrexp((E_newE_current).t)E_current=E_newsol_current=sol_newelsesol_new=sol_currentendendendt=t.a%控制參數(shù)t（溫度）減少為原來的a倍enddisp(最優(yōu)解為：)disp(sol_best)disp(最短距離：)disp(E_best)figure()set(gcfNamemonituihuosol_bestColr

6、)N=length(sol_best)scatter(codinates(:1)codinates(:2)50filled)holdonplot([codinates(sol_best(1)1)codinates(sol_best(N)1)][codinates(sol_best(1)2)codinates(sol_best(N)2)])set(gcaColg)holdonfi=2:Nplot([codinates(sol_best(i