dss.m

来自「一种新的时频分析方法的matlab源程序。」· M 代码 · 共 48 行

function z=dss(nt,dt)
%    z=dss(nt,dt):
%
%    Function to analyze the degree of stationarity of data nt(n,m) by
%    calculating the variance for each n, where n specifies the number 
%    of frequency values, and the length of time series.  
%    The procedure includes filtering of the input data.
%
%    Input-
%	nt	- 2-D matrix nt(n,m) that specifies the HHT spectrum
% 	dt	- positive integer number of points that specify 
%		  the smoothing length
%    Output-
%	z	- vector that specifies the degree of stationarity 
%		  for each frequency value
%
%    Norden Huang (NASA GSFC)	Initial
%    Jelena Marshak (NASA GSFC)	Nov.7, 2003 Modified and edited
%				(added mean value check for zero).
%
%    Notes-
%    An input matrix can be generated by calling a function that
%    generates the spectrum, for example 'nsp()' or 'nspab()'.
%    If mean value is zero, its components are assumed to be near zero.

%---- Get dimensions (number of frequency and time values)
ii=size(nt);
jf1=ii(1);

%---- Calculate the mean
ms=sum(nt,2);
ms=ms/ii(2);

%---- Filter, normalize and calculate the variance
dt1=ones(dt)/dt;
for i=1:jf1
   nt(i,:)=filtfilt(dt1,1,nt(i,:));
   if(ms(i) ~=0)
   nt(i,:)=nt(i,:)/ms(i);
   end
end
nt=1-nt;
nt=nt.*nt;
z=sum(nt,2);
z=z/ii(2);
z(1)=0.;