|
楼主 |
发表于 2007-8-27 10:01
|
显示全部楼层
function [s,delt_s,s_cor]=C_CMethod(data)
%this function calculate time delay and embedding demension with C-C
%Method,which proved by H.S.Kim
%skyhawk&flyinghawk
% %****************?????****************************
% clear all;
% data=load('C:\Documents and Settings\Administrator\??\a.txt');% ????
% %************************************************
data=[0.2483 0.1337 0.0993 0.1648 0.2303 0.2967 0.3504 0.2319 0.1627 0.0941 0.0523 0.0469 0.0944 0.0698 0.3154 0.4245 0.5124 0.3651 0.6215 0.1235 0.842 0.5124];
N=length(data)
max_d=2;%the maximum value of the time delay
sigma=std(data);%calcute standard deviation s_d
for t=1:max_d
t
s_t=0;
delt_s_s=0;
for m=2:5
s_t1=0;
for j=1:4
r=sigma*j/2;
data_d=disjoint(data,N,t);%将时间序列分解成t个不相交的时间序列
[ll,N_d]=size(data_d);
s_t3=0;
for i=1:t
i
Y=data_d(i,:);
C_1(i)=correlation_integral(Y,N_d,r);%计算C(1,N_d,r,t)
X=reconstitution(Y,N_d,m,t);%相空间重构
N_r=N_d-(m-1)*t;
C_I(i)=correlation_integral(X,N_r,r);%计算C(m,N_r,r,t)
s_t3=s_t3+(C_I(i)-C_1(i)^m);%对t个不相关的时间序列求和
end
s_t2(j)=s_t3/t;
s_t1=s_t1+s_t2(j);%对rj求和
end
delt_s_m(m)=max(s_t2)-min(s_t2);%求delt S(m,t)
delt_s_s=delt_s_s+delt_s_m(m);%delt S(m,t)对m求和
s_t0(m)=s_t1;
s_t=s_t+s_t0(m);%S对m求和
end
s(t)=s_t/16;
delt_s(t)=delt_s_s/4;
s_cor(t)=delt_s(t)+abs(s(t));
end
fid=fopen('result.txt','w');
fprintf(fid,'%f %f %f %f/n',t,s(t),delt_s(t),s_cor(t));
fclose(fid);
t=1:max_d;
plot(t,s,t,delt_s,'.',t,s_cor,'*')
这是C-C方法计算时间延迟的算法, 在计算的时候要注意你数据和max_d的关系 数据多了才能是max_d选择更大的值 |
|