萤火虫算法优化BP神经网络正交实验回归拟合，fa-bp神经网络正交实验回归拟合

%% this code for regression based on fa-bp %% initiate clc clear close all tic load maydata.mat input = bx(28::,1:4); output = bx(:,5); %% normlization [inputn,inputps] = mapminmax(input',-1,1); [outputn,outputps]=mapminmax(output',-1,1); n = randperm(length(inputn)); m = 70; Pn_train=inputn(:,n(1:m)); Tn_train=outputn(:,n(1:m)); Pn_test=inputn(:,n(m+1:end)); T_test=outputn(:,n(m+1:end)); % number of node inputnum=size(Pn_train,1); hiddennum=10; outputnum=1; % net=newff(Pn_train,Tn_train,hiddennum); net=newff(minmax(Pn_train),[10,1],{'logsig','tansig'},'traingdx');%tansig net.trainParam.epochs=200; net.trainParam.lr=0.1; net.trainParam.goal=0.00001; [net]=train(net,Pn_train,Tn_train); an=sim(net,Pn_test); test_simu=mapminmax('reverse',an,outputps); figure plot(test_simu,'b-o'); hold on; plot(output(n(m+1:end)),'k-*'); legend('预测数据','期望数据'); title('优化前') %% d=inputnum*hiddennum+hiddennum+hiddennum*outputnum+outputnum;% Lb=zeros(1,d); % lower Ub=ones(1,d); % upper u0=Lb+(Ub-Lb).*rand(1,d); para=[20 50 0.25 0.20 1]; % n=para(1); % number of fireflies % MaxGeneration=para(2); %number of pseudo time steps % ------------------------------------------------ % alpha=para(3); % Randomness 0--1 (highly random) % betamn=para(4); % minimum value of beta % gamma=para(5); % Absorption coefficient [nbest,trace]=ffa_mincon(u0,Lb,Ub,para,inputnum,hiddennum,outputnum,net,inputn,outputn); figure plot(trace) xlabel('迭代次数') ylabel('适应度值') title('萤火虫优化BP适应度曲线') x=nbest; %% net=newff(minmax(Pn_train),[10,1],{'logsig','purelin'},'traingdx');% 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; net.trainParam.epochs=2000; net.trainParam.lr=0.1; net.trainParam.goal=0.001; [net,per2]=train(net,Pn_train,Tn_train); % 反归一化 an=sim(net,Pn_test); test_simu=mapminmax('reverse',an,outputps); T_test=mapminmax('reverse',T_test,outputps); error=test_simu-T_test; E=mean(abs(error./T_test)) figure plot(test_simu,'r-s') hold on; plot(T_test,'k-*') title('萤火虫优化bp后的结果','fontsize',12) legend('预测数据','期望数据') xlabel('样本') ylabel('y') toc