回归预测 | MATLAB实现PSO-BP多输入多输出

回归预测 | MATLAB实现PSO-BP多输入多输出

本程序基于MATLAB实现PSO-BP多输入多输出回归型预测，3个输入，2个输出。

主程序代码如下：

%% 清除工作空间中的变量和图形
clc
clear
%% 构造样本集
load dataOTH input output
%%节点个数
inputnum=3;
hiddennum=6;
outputnum=2;
numsum=inputnum*hiddennum+hiddennum+hiddennum*outputnum+outputnum;
%划分训练、测试样本
input_train=input(1:72,:)';
input_test=input(73:108,:)';
output_train=output(1:72,1:2)';
output_test=output(73:108,1:2)';
%数据归一化
[inputn,inputps]=mapminmax(input_train);
[outputn,outputps]=mapminmax(output_train);
%%构建网络
net=newff(inputn,outputn,hiddennum);
%参数初始化
%粒子群算法中的两个参数
c1 = 1.49445;
c2 = 1.49445;
%进化次数 
maxgen=20;   
%粒子规模
sizepop=20;   
%粒子和速度最大最小值
Vmax=1;
Vmin=-1;
popmax=5;
popmin=-5;
for i=1:sizepop
    pop(i,:)=5*rands(1,numsum);
    V(i,:)=rands(1,numsum);
    fitness(i)=fun(pop(i,:),inputnum,hiddennum,outputnum,net,inputn,outputn);
end
%寻找粒子极值和群体极值
[bestfitness,bestindex]=min(fitness);
%全局最佳
zbest=pop(bestindex,:);   
%个体最佳
gbest=pop;    
%个体最佳适应度值
fitnessgbest=fitness;   
%全局最佳适应度值
fitnesszbest=bestfitness;   
%% 迭代寻优
for i=1:maxgen
    i; 
    for j=1:sizepop
        %速度更新
        V(j,:) = V(j,:) + c1*rand*(gbest(j,:) - pop(j,:)) + c2*rand*(zbest - pop(j,:));
        V(j,find(V(j,:)>Vmax))=Vmax;
        V(j,find(V(j,:)popmax))=popmax;
        pop(j,find(pop(j,:)0.95
            pop(j,pos)=5*rands(1,1);
        end
        %适应度值
        fitness(j)=fun(pop(j,:),inputnum,hiddennum,outputnum,net,inputn,outputn);
    end
    for j=1:sizepop
    %群体最优更新
    if fitness(j) < fitnessgbest(j)
        gbest(j,:) = pop(j,:);
        fitnessgbest(j) = fitness(j);
    end
    %群体最优更新 
    if fitness(j) < fitnesszbest
        zbest = pop(j,:);
        fitnesszbest = fitness(j);
    end
    end
    yy(i)=fitnesszbest;    
end
x=zbest;
%% 
% %
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}=reshape(B2,outputnum,1);
%% 
net.trainParam.epochs=1000;
net.trainParam.lr=0.9;
net.trainParam.goal=0.00001;
net.trainParam.max_fail=6;
net.trainParam.min_grad=1e-15;
net.trainParam.showWindow=1;
net.trainFcn='trainlm';
[net,per2]=train(net,inputn,outputn);
inputn_test=mapminmax('apply',input_test,inputps);
an=sim(net,inputn_test);
test_simu=mapminmax('reverse',an,outputps);
error=test_simu-output_test;
variance=error./output_test;
mseindepth=mse(error(1,:));
avgerrorindepth=mean(error(1,:));
mseinra1=mse(error(2,:));
avgerrorinra1=mean(error(2,:));
%% 数据可视化
%% 
figure(1)
plot(yy,':o','Color',[0 0 180]./255,'linewidth',0.8,'Markersize',4,'MarkerFaceColor',[0 0 180]./255)
title('适应度曲线','fontsize',12)
xlabel('进化代数','fontsize',12);ylabel('Fitness','fontsize',12);
figure(2)
plot(test_simu(1,:),'r:o','Color',[255 0 0]./255,'linewidth',0.8,'Markersize',4,'MarkerFaceColor',[255 0 0]./255)
hold on
plot(output_test(1,:),'k-s','Color',[0 0 0]./255,'linewidth',0.8,'Markersize',5,'MarkerFaceColor',[0 0 0]./255);
legend('预测值','实际值')
title('BP神经网络预测1','fontsize',12)
ylabel('预测值','fontsize',12)
xlabel('样本值','fontsize',12);
figure(3)
plot(test_simu(2,:),'b:o','Color',[0 0 255]./255,'linewidth',0.8,'Markersize',4,'MarkerFaceColor',[0 0 255]./255)
hold on
plot(output_test(2,:),'k-s','Color',[0 0 0]./255,'linewidth',0.8,'Markersize',5,'MarkerFaceColor',[0 0 0]./255);
legend('预测值','实际值')
title('BP神经网络预测2','fontsize',12)
ylabel('预测值','fontsize',12)
xlabel('样本值','fontsize',12);
%% 误差分析
figure(4)
plot(variance(1,:)*100,'b-o','Color',[255 0 255]./255,'linewidth',0.8,'Markersize',4,'MarkerFaceColor',[255 0 255]./255);
title('预测相对误差1','fontsize',12)
ylabel('百分比误差(%)','fontsize',12)
xlabel('测试样本','fontsize',12);
figure(5)
plot(variance(2,:)*100,'b-o','Color',[180 60 0]./255,'linewidth',0.8,'Markersize',4,'MarkerFaceColor',[180 60 0]./255);
title('预测相对误差2','fontsize',12)
ylabel('百分比误差(%)','fontsize',12)
xlabel('测试样本','fontsize',12);

函数子文件如下：

function error = fun(x,inputnum,hiddennum,outputnum,net,inputn,outputn)
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=newff(inputn,outputn,hiddennum);
%Construction of BP network
net.trainParam.epochs=3000;
net.trainParam.lr=0.1;
net.trainParam.mc=0.9;
net.trainFcn='trainlm';
net.trainParam.goal=0.00001;
net.trainParam.show=50;
net.trainParam.showWindow=0;
%BP network training
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}=reshape(B2,outputnum,1);
net=train(net,inputn,outputn);
an=sim(net,inputn);
error=sum(abs(an(:)-outputn(:)));

运行结果如下：

• 建模不易，需要定制模型的话有偿咨询，需要的同学添加Q~Q【1153460737】交流，记得备注。
• 也欢迎关注公=众=号【交通专业小明同学】，一起学习，一起进步。
• 程序和数据下载地址参见本人主页资源。