能源行业(电力石油煤炭)

开发平台：

Matlab

test_sample_my.asv：源码内容

							%generate train date&test data
%this sample is from "Nonlinear process monitoring using kernel principal component analysis"
e11=normrnd(0,0.01,1,100);
e21=normrnd(0,0.01,1,100);
e31=normrnd(0,0.01,1,100);
t1=rand(1,100)*2;
x11=t1+e11;
for i=1:100
    x21(i)=2*t1(i)^2-3*t1(i)+e21(i);
end
for i=1:100
    x31(i)=-t1(i)^3+3*t1(i)^2+e31(i);
end
trian=[x11;x21;x31];
e1=normrnd(0,0.01,1,300);
e2=normrnd(0,0.01,1,300);
e3=normrnd(0,0.01,1,300);
t=rand(1,300)*2;
x1=t+e1;
for i=1:300
    x2(i)=2*t(i)^2-3*t(i)+e2(i);
end
for i=1:300
    x3(i)=-t(i)^3+3*t(i)^2+e3(i);
end
for i=101:250
    x2(i)=x2(i)-0.4;
end
tes=[x1;x2;x3];
%standard the data
[tr,meanp,stdp] = prestd(trian);
te = trastd(tes,meanp,stdp);
tr=tr';
te=te';
trd=data(tr);
ted=data(te)
[r,a]=train(kpca({kernel('rbf',1.581)}),trd);
d=test(a,ted);
val=a.e_val;
vec=a.e_vec;
dvec=d.X;
s=inv(diag(val(1:10)));
for i=1:300
  tsquare(i)=dvec(i,1:10)*s*dvec(i,1:10)';
end
for i=1:300
   spe(i)=dvec(i,10)^2;
   for j=11:70
        spe(i)=spe(i)+dvec(i,j)^2;
    end
end
%cn=kpcak(te,tr,vec(:,1:4),s);

						