MATLAB-BP神经网络算法实现

%% I. 清空环境变量 % clear all clc %% II. 绘制函数曲线 x = 0:0.01:9; y = x + 10*sin(5*x)+7*cos(4*x); % x = 0:0.01:8; % y = fitnessVal = 2 * exp(-x) * sin(x)); figure plot(x, y) xlabel('自变量') ylabel('因变量') %title('y = 2 * exp(-x) *sin(x)') title('x + 10*sin(5*x)+7*cos(4*x)') %% III. 初始化种群 initPop = initializega(50,[0 9],'fitness'); %% IV. 遗传算法优化 [x endPop bpop trace] = ga([0 9],'fitness',[],initPop,[1e-6 1 1],'maxGenTerm',25,... 'normGeomSelect',0.08,'arithXover',2,'nonUnifMutation',[2 25 3]); %% V. 输出最优解并绘制最优点 x hold on plot (endPop(:,1),1/endPop(:,2)-20,'ro') %% VI. 绘制迭代进化曲线 figure(2) plot(trace(:,1),trace(:,3),'b:') hold on plot(trace(:,1),trace(:,2),'r-') xlabel('Generation'); ylabel('Fittness'); legend('Mean Fitness', 'Best Fitness')