Prev: image processing
Next: Fuzzy logic toolbox
From: Agus Nugroho on 3 Jun 2010 11:19
Dear All,

I would like to ask you in such problem that i am facing now, since i am still beginner in the Matlab programming. Currently, I have signal with length of 1000 sample and I would like to devide it to be 10 segments so each of them will have 100 sample. For each segment I would like to implement FFT and windowed function. And finally, I want to calculate the average FFT for these 10 segments.

I have FFT program but it i is used for calculating for FFT of 1000 samples. I have tried to use "for" function but it seems useless for me. Would you help me by correcting my sample signal below to fulfill my objective? Thank you in advance

----------------------------------------------------------------------------------------
Fs=100;
Ts=1/Fs;
t=0:Ts:1000-Ts;
n=length(t);
y=2*sin(2*pi*0.5*t)+5*sin(2*pi*0.2*t)+10*sin(2*pi*0.1*t)+20*sin(2*pi*0.05*t);
figure(1)
plot(t,y)

windowHann=window(@hann,n).';
hannWindowFigure=figure;
plot(windowHann);

windowedSignal=windowHann.*y;
windowedSignalPlot=figure;
plot(t,windowedSignal)

[Yk,k]=positiveFFT(windowedSignal,Fs);
Yk=Yk*2;
fftWindowedSignalLinear=figure;
stem(k,2*abs(Yk));

xlabel('Spatial frequency (m-1)')
ylabel('|Y(k)|')

------------------------------------------------------------
From: Agus Nugroho on 3 Jun 2010 12:50
Positive FFT is function with :
function[YfreqD,freqRng]=positiveFFT(y,Fs)
N=length(y);
k=0:N-1;
T=N/Fs;
freqRng=k/T;
YfreqD=fft(y)/N;
cutOff=ceil(N/2);
YfreqD=YfreqD(1:cutOff);
freqRng=freqRng(1:cutOff);
From: TideMan on 3 Jun 2010 16:17
On Jun 4, 4:50 am, "Agus Nugroho" <agus_nugroh...(a)yahoo.com> wrote:
> Positive FFT is function with :
> function[YfreqD,freqRng]=positiveFFT(y,Fs)
> N=length(y);
> k=0:N-1;
> T=N/Fs;
> freqRng=k/T;
> YfreqD=fft(y)/N;
> cutOff=ceil(N/2);
> YfreqD=YfreqD(1:cutOff);
> freqRng=freqRng(1:cutOff);

This all looks pretty good to me.
The only thing I'd change is the range: YfreqD(2:cutOff) and
freqRng(2:cutOff).
This excludes the DC (zero frequency) component.

As for ensemble averaging, you just need to split your signal using
reshape:
y1=reshape(y,100,10);
then loop through each column applying the window and positiveFFTy to
each.
You need to have a go at this and come back to us if you have a
problem.
From: Agus Nugroho on 4 Jun 2010 06:24
Dear Dr TideMan,

Actually I have arround 30 segments so its little bit difficult to examine one by one. However, I tried to make the function "for", but it seems not working well. May you see what is wrong in my function sir? I am sorry since I just new in this field so perhaps the logarithms is mistakes.

------------------------------------------------------------------------------

fo=5;
Fs=100;
Ts=1/Fs;
t=0:Ts:0.9-Ts;
N=length(t);
y=2*sin(2*pi*fo*t);

Nseg= 10; %Number of segments
n=floor(N/Nseg); %Nu. of samples per segment

Y_average=zeros(1,n);
for I=1:Nseg %number of segment
w=hann(n);
y1=y(1:n,I); %signal from segment 1 until segment 10
windowedSignal=(w(1:n,I).*y1; %windowoded signal for each segment

Y=fft(windowedSignal,n); %fft for each signal segment
end
%averages FFT
Y_average=Y_average+Y.*conj(Y);
stem(abs(Y_average));

----------------------------------------------------------

Thank you very much
From: Agus Nugroho on 7 Jun 2010 05:15
Dear Dr. Mulgor,

I have modify the algorithm as suggested from you in the previous email and its work.
You are so mch helpful

Thank you for your kindness.

Best regards,

Agus
 | 
Pages: 1
Prev: image processing
Next: Fuzzy logic toolbox