RSGABP神经网络故障诊断

% % 单层BP神经网络 % %% 此程序matlab编程实现的BP神经网络 % % 清空环境变量 % clear all % clc % input=xlsread('D:\BP神经网络\程序\17000数据.xlsx') % save data input % output=xlsread('D:\BP神经网络\程序\卷曲温度.xlsx') % save data output % in=input(1:17100,1:10)'; % out=output(1:17100,1); % %%第一步 读取数据 % input=in; %载入输入数据 % output=out; %载入输出数据 % % %% 第二步 设置训练数据和预测数据 % input_train = in(1:10,1:16000); % output_train =out(1:16000,1)'; % input_test = in(1:10,16001:17100); % output_test =out(16001:17100,1)'; % %% 第三本 训练样本数据归一化 % [inputn,inputps]=mapminmax(input_train);%归一化到[-1,1]之间，inputps用来作下一次同样的归一化 % [outputn,outputps]=mapminmax(output_train); % M=size(input,10); %输入节点个数 % N=size(output,1); %输出节点个数 % % n=6; %隐形节点个数 % lr1=0.01; %学习概率 % lr2=0.001; %学习概率 % maxgen=100; %迭代次数 % % %权值初始化 % Wjk=randn(n,M);Wjk_1=Wjk;Wjk_2=Wjk_1; % Wij=randn(N,n);Wij_1=Wij;Wij_2=Wij_1; % a=randn(1,n);a_1=a;a_2=a_1; % b=randn(1,n);b_1=b;b_2=b_1; % % %节点初始化 % y=zeros(1,N); % net=zeros(1,n); % net_ab=zeros(1,n); % % %权值学习增量初始化 % d_Wjk=zeros(n,M); % d_Wij=zeros(N,n); % d_a=zeros(1,n); % d_b=zeros(1,n); % % %%网络训练 % for i=1:maxgen % % %误差累计 % error(i)=0; % % % 循环训练 % for kk=1:size(input,1) % x=inputn(kk,:); % yqw=outputn(kk,:); % % for j=1:n % for k=1:M % net(j)=net(j)+Wjk(j,k)*x(k); % net_ab(j)=(net(j)-b(j))/a(j); % end % temp=mymorlet(net_ab(j)); % for k=1:N % y=y+Wij(k,j)*temp; %小波函数 % end % end % error(i)=error(i)+sum(abs(yqw-y)); % for j=1:n % %计算d_Wij % temp=mymorlet(net_ab(j)); % for k=1:N % d_Wij(k,j)=d_Wij(k,j)-(yqw(k)-y(k))*temp; % end % %计算d_Wjk % temp=d_mymorlet(net_ab(j)); % for k=1:M % for l=1:N % d_Wjk(j,k)=d_Wjk(j,k)+(yqw(l)-y(l))*Wij(l,j) ; % end % d_Wjk(j,k)=-d_Wjk(j,k)*temp*x(k)/a(j); % end % %计算d_b % for k=1:N % d_b(j)=d_b(j)+(yqw(k)-y(k))*Wij(k,j); % end % d_b(j)=d_b(j)*temp/a(j); % %计算d_a % for k=1:N % d_a(j)=d_a(j)+(yqw(k)-y(k))*Wij(k,j); % end % d_a(j)=d_a(j)*temp*((net(j)-b(j))/b(j))/a(j); % end % end % end % % % 第七步 测试样本归一化 % inputn_test=mapminmax('apply',input_test,inputps);% 对样本数据进行归一化 % % %%网络预测 % for i=1:92 % x_test=x(i,:); % % for j=1:1:n % for k=1:1:M % net(j)=net(j)+Wjk(j,k)*x_test(k); % net_ab(j)=(net(j)-b(j))/a(j); % end % temp=mymorlet(net_ab(j)); % for k=1:N % y(k)=y(k)+Wij(k,j)*temp ; % end % end % % yuce(i)=y(k); % y=zeros(1,N); % net=zeros(1,n); % net_ab=zeros(1,n); % end % %预测输出反归一化 % ynn=mapminmax('reverse',yuce,outputps); %% 该代码为基于小波神经网络的预测代码 %% 网络参数配置 % clear,clc % DATA=xlsread('testdata.xlsx'); % data=DATA(:,1:end-12); % % data1=DATA(end-11:end,5:6)-100; % fore=24; % test=12; % obs=12; function [InsampleReal,InsamplePredict,RMSE,OutsamplePredict,AdsustedR,adftestpValue,archtestpValue,dwtestpValue,R2testpValue,FtestpValue,para1,para2,para3,para4]=WANN(data,obs,test,fore,step) adftestpValue=[]; archtestpValue=[]; dwtestpValue=[]; R2testpValue=[]; FtestpValue=[]; data=data'; code=data(1,:); data(1,:)=[]; xunlian=length(data(:,1))-test-obs; if xunlian<=0 disp('数据长度不足') InsampleReal=[];InsamplePredict=[];RMSE=[];OutsamplePredict=[]; return end for u=1:length(data(1,:)) % u=1; % test=12; shuju=data(:,u); aa=shuju(1:end-test,:); real(:,u)=shuju(end-test+1:end,:); % obs=x; yuce=fore; % yuce=12; % obs=18; for i=1:xunlian inputn(:,i)=aa(i:(obs-1+i),1); end outputn=aa(obs+1:xunlian+obs,1); outputn=outputn'; P=inputn'; T=outputn'; %% 输入输出数据归一化 [P,inputps]=mapminmax(P'); [T,outputps]=mapminmax(T'); input=P'; output=T'; %% M=size(input,2); %输入节点个数 N=size(output,2); %输出节点个数 %18 n=7; %隐形节点个数 lr1=0.005; %学习概率 lr2=0.005; %学习概率 maxgen=100; %迭代次数 seed=2; randn('state',seed); %% if xunlian>12 numB=12; else numB=xunlian; end % for flag=6;%6:numB-1 % n=flag; % %权值初始化 % Wjk=randn(n,M);Wjk_1=Wjk;Wjk_2=Wjk_1; % Wij=randn(N,n);Wij_1=Wij;Wij_2=Wij_1; % a=randn(1,n);a_1=a;a_2=a_1; % b=randn(1,n);b_1=b;b_2=b_1; %节点初始化 y=zeros(1,N); % y=output; net=zeros(1,n); net_ab=zeros(1,n); %权值学习增量初始化 d_Wjk=zeros(n,M); d_Wij=zeros(N,n); d_a=zeros(1,n); d_b=zeros(1,n); %% 输入输出数据归一化 %权值初始化 Wjk=randn(n,M);Wjk_1=Wjk;Wjk_2=Wjk_1; Wij=randn(N,n);Wij_1=Wij;Wij_2=Wij_1; a=randn(1,n);a_1=a;a_2=a_1; b=randn(1,n);b_1=b;b_2=b_1; error=zeros(1,maxgen); %% 网络训练 for s=1:maxgen %误差累计 error(s)=0; % 循环训练 xx=input; yqw=output; for yb=1:length(xx(:,1)) x=xx(yb,:); for j=1:n for k=1:M net(j)=net(j)+Wjk(j,k)*x(k); net_ab(j)=(net(j)-b(j))/a(j); end temp=mymorlet(net_ab(j)); end for j=1:n for k=1:N y(k)=y(k)+Wij(k,j)*temp; %小波函数 end end ycc=y(k); % end %计算误差 s=1; error(s)=error(s)+sum(abs(yqw(yb,1)-ycc)); %权值调整 for j=1:n %计算d_Wij temp=mymorlet(net_ab(j)); for k=1:N d_Wij(k,j)=d_Wij(k,j)-sum(yqw(:,1)-ycc)*temp; end %计算d_Wjk temp=d_mymorlet(net_ab(j)); for k=1:M for l=1:N d_Wjk(j,k)=d_Wjk(j,k)+sum(yqw(:,1)-ycc)*Wij(l,j) ; end d_Wjk(j,k)=-d_Wjk(j,k)*temp*x(k)/a(j); end %计算d_b for k=1:N d_b(j)=d_b(j)+sum(yqw(:,1)-ycc)*Wij(k,j); end d_b(j)=d_b(j)*temp/a(j); %计算d_a for k=1:N d_a(j)=d_a(j)+sum(yqw(:,1)-ycc)*Wij(k,j); end d_a(j)=d_a(j)*temp*((net(j)-b(j))/b(j))/a(j); end %权值参数更新 Wij=Wij-lr1*d_Wij; Wjk=Wjk-lr1*d_Wjk; b=b-lr2*d_b; a=a-lr2*d_a; d_Wjk=zeros(n,M); d_Wij=zeros(N,n); d_a=zeros(1,n); d_b=zeros(1,n); y=zeros(1,N); net=zeros(1,n); net_ab=zeros(1,n); Wjk_1=Wjk;Wjk_2=Wjk_1; Wij_1=Wij;Wij_2=Wij_1; a_1=a;a_2=a_1; b_1=b;b_2=b_1; end end para1(:,:,u)=Wjk; para2(:,:,u)=Wij; para3(:,u)=a; para4(:,u)=b; %% %% 网络预测 %预测输入归一化 % outputtest=aa(xunlian+1:end,1); % outputtest=outputtest'; inputtest=aa(xunlian+1:end,1); x_testn=inputtest'; s=fore+test; %yuce=zeros(1,s); z=zeros(1,N); %网络预测 x_test=mapminmax('apply',x_testn',inputps); x_test=x_test'; for i=1:s % i=2; %预测输入归一化 for j=1:1:n for k=1:1:M net(j)=net(j)+Wjk(j,k)*x_test(k); net_ab(j)=(net(j)-b_1(j))/a_1(j); end temp=mymorlet(net_ab(j)); for k=1:N z0=z(k)+Wij(k,j)*temp ; end end yuce(i,u)=z0; z0=0; z=zeros(1,N); net=zeros(1,n); net_ab=zeros(1,n); x_test=[x_test(2:end) yuce(i,u)]; end if step == 0 yuce(:,u) = yuce(:,u); else Randomerror(u,:)=normrnd(0,sqrt(var(data(:,u))),1,step); yuce(:,u) = yuce(:,u)+mean(Randomerror(u,:)); end %预测输出反归一化 ynn(:,u)=mapminmax('reverse',yuce(:,u),outputps);input=[];output=[]; InsampleReal(:,u)=real(:,u); ynn(isnan(ynn)==1)=0; ynn(isinf(ynn)==1)=0; InsamplePredict(:,u)=ynn(1:test,u); OutsamplePredict(:,u)=ynn(test+1:test+fore,u); RMSE(u)=sqrt(sum((real(:,u)-ynn(1:test,u)).^2)/test); AdsustedR(u)=sum((mean(real(:,u))-yn