遺傳算法優(yōu)化BP神經(jīng)網(wǎng)絡實現(xiàn)代碼

1、%讀取數(shù)據(jù)data=xlsread('data.xls'）；%訓練預測數(shù)據(jù)data_train=data(1:113,:);data_test=data(118:123,:);input_train=data_train(:,1:9),;output_train=data_train(:,10)'input_test=data_test(:,1:9),;output_test=data_test(:,10)'%數(shù)據(jù)歸一化inputn,mininput,maxinput,outputn,minoutput,maxoutput=premnmx(input_train

2、,output_train);%寸p和t進行字標準化預處理net=newff(minmax(inputn),10,1,'tansig','purelin','trainlm');net.trainParam.epochs=100;net.trainParam.lr=0.1;net.trainParam.goal=0.00001;%net.trainParam.show=NaN%網(wǎng)絡訓練net=train(net,inputn,outputn);%數(shù)據(jù)歸一化inputn_test=tramnmx(input_test,mininput,maxinp

3、ut);an=sim(net,inputn);test_simu=postmnmx(an,minoutput,maxoutput);error=test_simu-output_train;plot(error)k=error./output_trainfunctionret=Cross(pcross,lenchrom,chrom,sizepop,bound)%本函數(shù)完成交叉操作%pcorssinput:交叉概率%lenchrominput:染色體的長度%chrominput:染色體群%sizepopinput:種群規(guī)模%retoutput:交叉后的染色體fori=1:sizepop%每一輪f

4、or循環(huán)中，可能會進行一次交叉操作，染色體是隨機選擇的，交叉位置也是隨機選擇的，但該輪for循環(huán)中是否進行交叉操作則由交叉概率決定（continue控制）%隨機選擇兩個染色體進行交叉pick=rand(1,2);whileprod(pick)=0pick=rand(1,2);endindex=ceil(pick.*sizepop);%交叉概率決定是否進行交叉pick=rand;whilepick=0pick=rand;endifpick>pcrosscontinue;endflag=0;whileflag=0%隨機選擇交叉位pick=rand;whilepick=0pick=rand;%

5、隨機選擇進行交叉的位置，即選擇第幾個變%交叉結束%險驗染色體1的可行性%險驗染色體2的可行性endpos=ceil(pick.*sum(lenchrom);量進行交叉，注意：兩個染色體交叉的位置相同pick=rand;%5叉開始v1=chrom(index(1),pos);v2=chrom(index(2),pos);chrom(index(1),pos)=pick*v2+(1-pick)*v1;chrom(index(2),pos)=pick*v1+(1-pick)*v2;flag1=test(lenchrom,bound,chrom(index(1),:);flag2=test(lench

6、rom,bound,chrom(index(2),:);ifflag1*flag2=0flag=0;elseflag=1;end%如果兩個染色體不是都可行，則重新交叉endendret=chrom;%清空環(huán)境變量clcclear%網(wǎng)絡結構建立%讀取數(shù)據(jù)loaddatainputoutput%節(jié)點個數(shù)inputnum=2;hiddennum=5;outputnum=1;%訓練數(shù)據(jù)和預測數(shù)據(jù)input_train=input(1:1900,:)'input_test=input(1901:2000,:)'output_train=output(1:1900)'output_

7、test=output(1901:2000)'%選連樣本輸入輸出數(shù)據(jù)歸一化inputn,inputps=mapminmax(input_train);outputn,outputps=mapminmax(output_train);%勾建網(wǎng)絡net=newff(inputn,outputn,hiddennum);%遺傳算法參數(shù)初始化maxgen=10;%進化代數(shù)，即迭代次數(shù)sizepop=10;%中群規(guī)模pcross=0.3;%交叉概率選擇，0和1之間pmutation=0.1;做異概率選擇，0和1之間%節(jié)點總數(shù)numsum=inputnum*hiddennum+hiddennum+h

8、iddennum*outputnum+outputnum;lenchrom=ones(1,numsum);bound=-3*ones(numsum,1)3*ones(numsum,1);磁據(jù)范圍%種群初始化individuals=struct('fitness',zeros(1,sizepop),'chrom',);%等種群信息定義為一個結構體avgfitness=;%每一代種群的平均適應度bestfitness=;%每一代種群的最佳適應度bestchrom=;%適應度最好的染色體%初始化種群fori=1:sizepop%!機產(chǎn)生一個種群individuals.

9、chrom(i,:)=Code(lenchrom,bound);哪碼(binary和grey胖編碼結果為一個實數(shù)，float的編碼結果為一個實數(shù)向量)x=individuals.chrom(i,:);%十算適應度individuals.fitness(i)=fun(x,inputnum,hiddennum,outputnum,net,inputn,outputn);礎色體的適應度end%找最好的染色體bestfitnessbestindex=min(individuals.fitness);bestchrom=individuals.chrom(bestindex,:);%>好的染色體a

10、vgfitness=sum(individuals.fitness)/sizepop;礎色體的平均適應度%記錄每一代進化中最好的適應度和平均適應度trace=avgfitnessbestfitness;%迭代求解最佳初始閥值和權值%進化開始fori=1:maxgeni%選擇individuals=Select(individuals,sizepop);avgfitness=sum(individuals.fitness)/sizepop;伙叉individuals.chrom=Cross(pcross,lenchrom,individuals.chrom,sizepop,bound)；%變異i

11、ndividuals.chrom=Mutation(pmutation,lenchrom,individuals.chrom,sizepop,i,maxgen,bound);%計算適應度forj=1:sizepopx=individuals.chrom(j,:);%單碼individuals.fitness(j)=fun(x,inputnum,hiddennum,outputnum,net,inputn,outputn);end%找到最小和最大適應度的染色體及它們在種群中的位置newbestfitness,newbestindex=min(individuals.fitness);worest

12、fitness,worestindex=max(individuals.fitness);%代替上一次進化中最好的染色體ifbestfitness>newbestfitnessbestfitness=newbestfitness;bestchrom=individuals.chrom(newbestindex,:);endindividuals.chrom(worestindex,:)=bestchrom;individuals.fitness(worestindex)=bestfitness;avgfitness=sum(individuals.fitness)/sizepop;tra

13、ce=trace;avgfitnessbestfitness;%記錄每一代進化中最好的適應度和平均適應度end%遺傳算法結果分析figure(1)rc=size(trace);plot(1:r',trace(:,2),'b-');title('適應度曲線''終止代數(shù)='num2str(maxgen);xlabel('進化代數(shù));ylabel('適應度');legend('平均適應度，最佳適應度)；disp('適應度變量);x=bestchrom;%把最優(yōu)初始閥值權值賦予網(wǎng)絡預測%用遺傳算法優(yōu)化的BP

14、網(wǎng)絡進行值預測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.iw1,1=reshape(w1,hiddennum,inputnum);net.lw2,1=reshape(w2,outputnum,hiddennum);net.b1=reshape(B1,hiddennum,1);net.b2=B2;%BP網(wǎng)絡訓練%網(wǎng)絡進化參數(shù)net.trainParam.epochs=100;net.trainParam.lr=0.1;%net.trainPa