dfwkmc.m

其中有五个程序

                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %          程序运行过程中调用:dddddc,dddddl函数,这个函数计算在本聚类模式下的两个模糊数的距离          %
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function DFWKMC(k,mm,x,e,v)    %其中，e表示预设的阈值；k类，mm表示模糊关系数，x为待聚类的模糊数据矩阵，
                               %每一行表示一个模糊数.若每一个模糊数包含m个不同的数据单元,则要求x为一个n*(2m)的矩阵.其中n表示
                               %模糊数的个数,每一行中前m个表示该模糊数据的m个中心位置,后m个数据表示区间长。当然，
                               %这种模式也只能处理左右对称型的数据。其中，v表示这个计算模型下要求的参数，其范围应该是
                               %0<=v<=0.5
n=size(x);
m=n(2)/2;                      %求出模糊数据向量的规模，即其包含几个部分
n=n(1);                        %求出x的模糊数的个数n



                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %       判断类别数是不是大于数据个数      %
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
if k>n
    fprintf('ERROR! The number of clusters is greater than the number of the data, namely %d>%d. Please revise your cluster number!\n',k,n)
else


                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %                             声明变量空间                                                 
                              %    u：为初始关系矩阵
                              %    uu：为经优化后的关系矩阵
                              %    c：为数据的中心矩阵（这是一个矩阵，因为x为一个向量模糊数）                
                              %    l：为数据的区间长矩阵
                              %    C：为k类初始中心矩阵
                              %    L：为k类初始区间长矩阵
                              %    CC：为k类经优化后的中心矩阵
                              %    LL：k类经优化后的左区间长矩阵
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
u=ones(k,n)/n;
uu=ones(k,n);
c=ones(n,m);
l=ones(n,m);
C=ones(k,m);
CC=ones(k,m);
L=ones(k,m);
LL=ones(k,m);

                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %                         初始赋值
                              %   第一部分，将x分为中心和区间长，分别赋值给c，g，l和r
                              %   第二部分，选取k个聚类中心，区间长。不妨取为前k个元素，其中为了计算需要对中心做一个平移变换
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for i=1:n                         %第一部分
    c(i,:)=x(i,1:m);
    l(i,:)=x(i,m+1:2*m);
end
for i=1:k                         %第二部分
    C(i,:)=c(i,:)-0.1;
    L(i,:)=l(i,:);
end

                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %                  计算部分
                              %     第三部分，计算优化后的关系矩阵。其中用到的算法是根据论文中提供的方法。
                              %     第四部分，为计算优化后的聚类中心和区间长度矩阵。其中用到的算法是根据论文中提供的方法。
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
for j=1:n                         %第三部分
    d=ones(1,k);
    for i=1:k
        d(i)=1/(((1-v)^2*dddddc(x(j,:),[C(i,:),L(i,:)])^2+v^2*dddddl(x(j,:),[C(i,:),L(i,:)])^2)^(1/(mm-1)));
    end
    for i=1:k
        uu(i,j)=1/((((1-v)^2*dddddc(x(j,:),[C(i,:),L(i,:)])^2+v^2*dddddl(x(j,:),[C(i,:),L(i,:)])^2)^(1/(mm-1)))*sum(d));
    end
end
for i=1:k                         %第四部分
       CC(i,:)=0;
       LL(i,:)=0;
     for j=1:n
         CC(i,:)=uu(i,j)^mm*c(j,:)+CC(i,:);
         LL(i,:)=uu(i,j)^mm*l(j,:)+LL(i,:);
     end
     CC(i,:)=CC(i,:)/(sum(uu(i,:).^mm));
     LL(i,:)=LL(i,:)/(sum(uu(i,:).^mm));
end
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              % 计算v根据论文要求，若v大于0.5，则将其值修正为0.5
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
fz=ones(k,n);
for i=1:k
    for j=1:n
        fz(i,j)=dddddc(x(j,:),[CC(i,:),LL(i,:)])^2;
    end
end
fm=ones(k,n);
for i=1:k
    for j=1:n
        fm(i,j)=dddddc(x(j,:),[CC(i,:),LL(i,:)])^2+dddddl(x(j,:),[CC(i,:),LL(i,:)])^2;
    end
end
v=sum(sum((uu.^mm).*fz))/sum(sum((uu.^mm).*fm));
if v>0.5
    v=0.5;
end

                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %             进入循环项：若u和uu相应元素的差的最大值大于预设阈值e，则进入循环
                              %  将新算出的结果赋给就变量，重新计算相应的关系矩阵uu和聚类中心和区间矩阵，直到循环条件不被满足，跳出循环。
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
while max(max(abs(u-uu)))>e 
    u=uu;
    C=CC;
    L=LL;
    for j=1:n                         %第三部分
    d=ones(1,k);
    for i=1:k
        d(i)=1/(((1-v)^2*dddddc(x(j,:),[C(i,:),L(i,:)])^2+v^2*dddddl(x(j,:),[C(i,:),L(i,:)])^2)^(1/(mm-1)));
    end
    for i=1:k
        uu(i,j)=1/((((1-v)^2*dddddc(x(j,:),[C(i,:),L(i,:)])^2+v^2*dddddl(x(j,:),[C(i,:),L(i,:)])^2)^(1/(mm-1)))*sum(d));
    end
end
for i=1:k                         %第四部分
       CC(i,:)=0;
       LL(i,:)=0;
     for j=1:n
         CC(i,:)=uu(i,j)^mm*c(j,:)+CC(i,:);
         LL(i,:)=uu(i,j)^mm*l(j,:)+LL(i,:);
     end
     CC(i,:)=CC(i,:)/(sum(uu(i,:).^mm));
     LL(i,:)=LL(i,:)/(sum(uu(i,:).^mm));
end
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              % 计算v根据论文要求，若v大于0.5，则将其值修正为0.5
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
fz=ones(k,n);
for i=1:k
    for j=1:n
        fz(i,j)=dddddc(x(j,:),[CC(i,:),LL(i,:)])^2;
    end
end
fm=ones(k,n);
for i=1:k
    for j=1:n
        fm(i,j)=dddddc(x(j,:),[CC(i,:),LL(i,:)])^2+dddddl(x(j,:),[CC(i,:),LL(i,:)])^2;
    end
end
v=sum(sum((uu.^mm).*fz))/sum(sum((uu.^mm).*fm));
if v>0.5
    v=0.5;
end
end
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %    输出计算的结果，分别为聚类中心和关系矩阵
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
fprintf('The centers of the %d clustering is:\n',k)
CC
fprintf('The membership of the elements relating to the %d clusters is:\n',k)
uu=uu'
fprintf('The graph has shown the result of the clustering, in which the red point is the center of the clusters.\n')
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
                              %     画图
                              %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
hold on
if m==2
for i=1:n
    plot([c(i,1)],[c(i,2)],'*')
    plot([c(i,1)-l(i,1),c(i,1)+l(i,1),c(i,1)+l(i,1),c(i,1)-l(i,1),c(i,1)-l(i,1)],[c(i,2)-l(i,2),c(i,2)-l(i,2),c(i,2)+l(i,2),c(i,2)+l(i,2),c(i,2)-l(i,2)])
end
else 
    for i=1:n
    plot([c(i,1)],[c(i,2)],'*')

    end
end
for i=1:k
plot([CC(i,1)],[CC(i,2)],'Marker','o','LineStyle','none','Color',[1 0 0])
end
hold off
end