PSOBP汇率预测.zip_BP_BP优化_PSO_神经网络_预测算法

%% 基于粒子群算法优化BP神经网络的汇率预测 clear all clc warning off %% 导入数据 load exchange_rate.mat x = []; y = []; tr_len = 800; num_input = 10; for i = 1:length(X)-num_input x = [x; X(i:i+num_input-1)]; y = [y; X(i+num_input)]; end %训练集——800个样本 input_train = x(1:tr_len, :)'; output_train = y(1:tr_len)'; %测试集——52个样本 input_test = x(tr_len+1:end, :)'; output_test = y(tr_len+1:end)'; %% BP网络设置 %节点个数 [inputnum,N]=size(input_train);%输入节点数量 outputnum=size(output_train,1);%输出节点数量 hiddennum=6; %选连样本输入输出数据归一化 [inputn,inputps]=mapminmax(input_train); %构建网络 net=newff(inputn,output_train,hiddennum); %% GA算法参数初始化 nvar=inputnum*hiddennum+hiddennum+hiddennum*outputnum+outputnum; [outputn outputps]=mapminmax(output_train);% 归一化到【0 1】之间 %% PSO 参数 c1 = 2; c2 = 2; w=0.2; % Vmin = -1; Vmax = 1; maxiter = 10; % 迭代次数 sizepop = 10; % 种群数量 popmin=-ones(1,nvar); % 下界 popmax=ones(1,nvar); % 上界 % 初始化种群 for i=1:sizepop pop(i,:) =init_individual(popmin,popmax,nvar,1); % 随机初始化个体 fitness(i)= objfun_BP(pop(i,:),inputnum,hiddennum,outputnum,net,inputn,outputn); V(i,:)=Vmin+(Vmax-Vmin)*rand(1,nvar); end % 记录一组最优值 [bestfitness,bestindex]=min(fitness); zbest=pop(bestindex,:); % 全局最佳 gbest=pop; % 个体最佳 fitnessgbest=fitness; % 个体最佳适应度值 fitnesszbest=bestfitness; % 全局最佳适应度值 % 迭代寻优 trace=[]; for i=1:maxiter i for j=1:sizepop % % % 速度更新 V(j,:) = w*V(j,:) + c1*rand*(gbest(j,:) - pop(j,:)) +c2*rand*(zbest - pop(j,:)); for iV = 1:nvar V(j,iV) = min( Vmax,V(j,iV)); V(j,iV) = max( Vmin,V(j,iV)); end % 个体更新 pop(j,:) = pop(j,:) + V(j,:); for iV = 1:nvar pop(j,iV) = min( popmax(iV), pop(j,iV)); pop(j,iV) = max( popmin(iV), pop(j,iV)); end % 适应度更新 fitness(j) = objfun_BP(pop(j,:),inputnum,hiddennum,outputnum,net,inputn,outputn); % end % 比较 个体间比较 if fitness(j)<fitnessgbest(j) fitnessgbest(j) = fitness(j); gbest(j,:) = pop(j,:); end if fitness(j)<fitnesszbest fitnesszbest = fitness(j); zbest = pop(j,:); end end fitness_iter(i) = fitnesszbest; avgfit(i)=sum(fitness)/sizepop; end disp(zbest) figure plot(fitness_iter,'b-*','linewidth',2) xlabel('迭代次数') ylabel('适应度') grid on x=zbest; %% 把最优初始阀值权值赋予网络预测 %% 把最优初始阀值权值赋予网络预测 % %用PSO算法优化的BP网络进行值预测 w1=x(1:inputnum*hiddennum); B1=x(inputnum*hiddennum+1:inputnum*hiddennum+hiddennum); w2=x(inputnum*hiddennum+hiddennum+1:inputnum*hiddennum+hiddennum+hiddennum*outputnum); B2=x(inputnum*hiddennum+hiddennum+hiddennum*outputnum+1:inputnum*hiddennum+hiddennum+hiddennum*outputnum+outputnum); net.iw{1,1}=reshape(w1,hiddennum,inputnum); net.lw{2,1}=reshape(w2,outputnum,hiddennum); net.b{1}=reshape(B1,hiddennum,1); net.b{2}=B2; %% BP网络训练 %网络进化参数 net.trainParam.epochs=100; net.trainParam.lr=0.1; net.trainParam.mc = 0.8;%动量系数，[0 1]之间 net.trainParam.goal=0.001; %网络训练 net=train(net,inputn,outputn); %% BP训练集预测 BP_sim=sim(net,inputn); %网络输出反归一化 T_sim=mapminmax('reverse',BP_sim,outputps); % figure plot(1:length(output_train),output_train,'b-','linewidth',1) hold on plot(1:length(T_sim),T_sim,'r-.','linewidth',1) axis tight xlabel('训练样本','FontSize',12); ylabel('汇率','FontSize',12); legend('实际值','预测值'); string={'PSO-BP预测'} title(string); % %% 测试数据归一化 inputn_test=mapminmax('apply',input_test,inputps); % %预测输出 an=sim(net,inputn_test); BPsim=mapminmax('reverse',an,outputps); figure plot(1:length(output_test), output_test,'b-','linewidth',1) hold on plot(1:length(BPsim),BPsim,'r-.','linewidth',1) xlabel('测试样本','FontSize',12); ylabel('汇率','FontSize',12); axis tight legend('实际值','预测值'); string={'PSO-BP预测'} title(string); % % 评价 ae= abs(BPsim - output_test); rmse = (mean(ae.^2)).^0.5; mae = mean(ae); mape = mean(100*ae./BPsim); disp('预测结果评价指标：') disp(['RMSE = ', num2str(rmse)]) disp(['MAE = ', num2str(mae)]) disp(['MAPE = ', num2str(mape)])