How to get the fundamental frequency magnitude ?

This is what I used to find the frequency and amplitude of the first 100 ms ( 1537 samples ) of the signal. The fprintf statement at the end writes the maximum amplitude and its frequency to the Command Window. This is the only way I know of to do what you want:

filedat = which('Andre_signal.csv');
x = csvread(filedat);
ad = x(:,6);
lendat = size(x,1);
Fs=15360; NFFT=512;
Ts = 1/Fs;
Tv = linspace(0,length(ad)/Fs,length(ad));
Pls1 = find(Tv > 0.1, 1, 'first');              % Initial ‘pulse’ about 100 ms long
Fv = [0:length(ad)/2-1]*Fs/length(ad);
ad2 = ad(1:Pls1);
Tv2 = linspace(0,length(ad2)/Fs,length(ad2));   % Time vector
Fv2 = [0:length(ad2)/2-1]*Fs/length(ad2);       % Frequency vector
fftad2 = fft(ad2);
afftad2 = abs(fftad2(1:fix(length(ad2)/2)));
figure(1)
plot(Tv2, ad2)
grid
maxix2 = find(afftad2 == max(afftad2));         % Find index of maximum fft amplitude
FVmax2 = [2*afftad2(maxix2)/length(ad2)    Fv2(maxix2)]
figure(2)
plot(Fv2, 2*afftad2/length(ad2))
axis([0  350    ylim])
grid
fprintf(1,'\n\tMaximum amplitude = %13.5E at %.2f Hz\n\n', FVmax2)

Star Strider el 6 de Mayo de 2014

My pleasure!

You probably can. I used the entire column in one part of my code because I didn’t get sufficiently detailed FFT results with only 400 samples ( 26 ms ) of a 60 Hz signal. That was only about 1.56 cycles, not enough for a reliable analysis, and gave a peak frequency of about 75 Hz, obviously wrong in the context of a longer and more reliable signal, and an inaccurate amplitude.
Your signal had three components, an initial 100 ms = 1537 sample segment (that I used as Pls1), a second smaller-amplitude segment with the same frequency about as long, and a third segment that seemed to repeat the first 100 ms. (There also appeared to be switching transients. They appeared in the analysis of the entire signal, but I did not include them in Pls1. I ignored all but the first 100 ms in my analysis that I posted here.) That was a longer version of the first 400 samples you wanted to use, and gave better results. I called it ‘Pulse#1’ for lack of a better term, and abbreviated it to Pls1. The FFT of Pls1 gave good results. The FFT of the first 400 samples did not.

Andre el 7 de Mayo de 2014

Thank you very much.

Andre

Star Strider el 7 de Mayo de 2014

Again, my pleasure!

Iniciar sesión para comentar.

Más respuestas (0)

Iniciar sesión para responder a esta pregunta.

Categorías

Signal Processing Signal Processing Toolbox Spectral Analysis Spectral Measurements

Más información sobre Spectral Measurements en Help Center y File Exchange.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by