MATLAB神经网络原理与实例精解_matlab_随机神经网络

% sa_tsp.m % 用模拟退火算法求解TSP问题 %% 清理 close all clear,clc %% 定义数据,position是2行25列的矩阵 position = [1304,2312;3639,1315;4177,2244;3712,1399;3488,1535;3326,1556;... 3238,1229;4196,1044;4312,790;4386,570;3007,1970;2562,1756;... 2788,1491;2381,1676;1322,695;3715,1678;3918,2179;4061,2370;... 3394,2643;3439,3201;2935,3240;3140,3550;2545,2357;2778,2826;2360,2975]'; L = length(position); % 计算邻接矩阵dist 25*25 dist = zeros(L,L); for i=1:L for j=1:L if i==j continue; end dist(i,j) = sqrt((position(1,i)-position(1,j)).^2 + (position(2,i)-position(2,j)).^2); dist(j,i) = dist(i,j); end end tic %% 初始化 MAX_ITER = 2000; MAX_M = 20; lambda = 0.97; T0 = 100; rng(2); x0 = randperm(L); %% T=T0; iter = 1; x=x0; % 路径变量 xx=x0; % 每个路径 di=tsp_len(dist, x0); % 每个路径对应的距离 n = 1; % 路径计数 % 外循环 while iter <=MAX_ITER, % 内循环迭代器 m = 1; % 内循环 while m <= MAX_M % 产生新路径 newx = tsp_new_path(x); % 计算距离 oldl = tsp_len(dist,x); newl = tsp_len(dist,newx); if ( oldl > newl) % 如果新路径优于原路径，选择新路径作为下一状态 x=newx; xx(n+1,:)=x; di(n+1)=newl; n = n+1; else % 如果新路径比原路径差，则执行概率操作 tmp = rand; if tmp < exp(-(newl - oldl)/T) x=newx; xx(n+1,:)=x; di(n+1)=newl; n = n+1; end end m = m+1; % 内循环次数加1 end % 内循环 iter = iter+1; % 外循环次数加1 T = T*lambda; % 降温 end toc %% 计算最优值 [bestd,index] = min(di); bestx = xx(index,:); fprintf('共选择 %d 次路径\n', n); fprintf('最优解:\n'); disp(bestd); fprintf('最优路线:\n'); disp(bestx); %% 显示 % 显示路径图 figure; plot(position(1,:), position(2,:),'o'); hold on; for i=1:L-1 plot(position(1,bestx(i:i+1)), position(2,bestx(i:i+1))); end plot([position(1,bestx(L)),position(1,bestx(1))], [position(2,bestx(L)),position(2,bestx(1))]); title('TSP问题选择的最优路径'); hold off; % 显示所选择的路径变化曲线 figure; semilogx(1:n,di); title('路径长度的变化曲线');