Using Findpeaks to plot on a scatter plot and return the peak coordinates

I used a FFT to transform my data from the time domain to the frequency domain. I also filtered the data to create localized frequency spikes, I am trying to identify the amplitude and frequency of the spikes. but when I use findpeaks it will not plot on my existing plot or show the correct peaks. Do I need to call the findpeaks function by a variable name so that I can access a matrix with the peak coordinates?

I have attached my code and data for your views. Thanks so much for your ideas.

%% Universal Constants

Fs=125; %to be checked Frequency Spectrum (62.5)

dt=1/Fs;

NFFT=800; %fft length (1024) changes the filter average

OVERLAP = .75; % between 0 and 1 max ; buffer overlap = OVERLAP*NFFT

%% Load Files

WBR_signal=load('WB0430Right.txt');

%% Samples and Channels for size

[R_samples_WB,R_channels_WB]=size(WBR_signal);

%% Selected Channels

selected_channel_AX=3;

selected_channel_AY=4;

selected_channel_AZ=5;

selected_channel_GX=6;

selected_channel_GY=7;

selected_channel_GZ=8;

selected_channel_MX=9;

selected_channel_MY=10;

selected_channel_MZ=11;

%% Samples and Channels for size

%WorkBoot

%Right

W_signal_RAX=WBR_signal(:,selected_channel_AX);

W_signal_RAY=WBR_signal(:,selected_channel_AY);

W_signal_RAZ=WBR_signal(:,selected_channel_AZ);

W_signal_RGX=WBR_signal(:,selected_channel_GX);

W_signal_RGY=WBR_signal(:,selected_channel_GY);

W_signal_RGZ=WBR_signal(:,selected_channel_GZ);

W_signal_RMX=WBR_signal(:,selected_channel_MX);

W_signal_RMY=WBR_signal(:,selected_channel_MY);

W_signal_RMZ=WBR_signal(:,selected_channel_MZ);

%% Time

%Work Boot

WR_time=(0:R_samples_WB-1)*dt;

%% Myfft specification

%Work Boot

%right

[W_freq_RAX,W_sensor_spectrum_RAX] = myfft_peak(W_signal_RAX,Fs,NFFT,OVERLAP);

[W_freq_RAY,W_sensor_spectrum_RAY] = myfft_peak(W_signal_RAY,Fs,NFFT,OVERLAP);

[W_freq_RAZ,W_sensor_spectrum_RAZ] = myfft_peak(W_signal_RAZ,Fs,NFFT,OVERLAP);

[W_freq_RGX,W_sensor_spectrum_RGX] = myfft_peak(W_signal_RGX,Fs,NFFT,OVERLAP);

[W_freq_RGY,W_sensor_spectrum_RGY] = myfft_peak(W_signal_RGY,Fs,NFFT,OVERLAP);

[W_freq_RGZ,W_sensor_spectrum_RGZ] = myfft_peak(W_signal_LGZ,Fs,NFFT,OVERLAP);

[W_freq_RMX,W_sensor_spectrum_RMX] = myfft_peak(W_signal_RMX,Fs,NFFT,OVERLAP);

[W_freq_RMY,W_sensor_spectrum_RMY] = myfft_peak(W_signal_RMY,Fs,NFFT,OVERLAP);

[W_freq_RMZ,W_sensor_spectrum_RMZ] = myfft_peak(W_signal_RMZ,Fs,NFFT,OVERLAP);

%% Figures

figure

plot(W_freq_RMZ,W_sensor_spectrum_RMZ/12.63)

xlabel('Frequency (Hz)')

ylabel('Amplitude')

ylim([0,1])

%Find Peaks

x=W_freq_RMZ;

y=W_sensor_spectrum_RMZ/12.63;

[peaks,locs]=findpeaks(y,x);

figure

z=plot(x,y,'b');

grid on

xlabel('Frequency (Hz)')

ylabel('Amplitude')

title('Peaks')

axis tight

hold on

scatter(peaks,locs,'^r');

%% Function

function [freq_vector,fft_spectrum] = myfft_peak(signal, Fs, nfft, Overlap)

% FFT peak spectrum of signal (example sinus amplitude 1 = 0 dB after fft).

% Linear averaging

% signal - input signal,

% Fs - Sampling frequency (Hz).

% nfft - FFT window size

% Overlap - buffer percentage of overlap % (between 0 and 0.95)

[samples,channels] = size(signal);

% fill signal with zeros if its length is lower than nfft

if samples<nfft

s_tmp = zeros(nfft,channels);

s_tmp((1:samples),:) = signal;

signal = s_tmp;

samples = nfft;

end

% window : hanning

window = hanning(nfft);

window = window(:);

% compute fft with overlap

offset = fix((1-Overlap)*nfft);

spectnum = 1+ fix((samples-nfft)/offset); % Number of windows

% % for info is equivalent to :

% noverlap = Overlap*nfft;

% spectnum = fix((samples-noverlap)/(nfft-noverlap)); % Number of windows

% main loop

fft_spectrum = 0;

for i=1:spectnum

start = (i-1)*offset;

sw = signal((1+start):(start+nfft),:).*(window*ones(1,channels));

fft_spectrum = fft_spectrum + (abs(fft(sw))*4/nfft); % X=fft(x.*hanning(N))*4/N; % hanning only

end

fft_spectrum = fft_spectrum/spectnum; % to do linear averaging scaling

% one sidded fft spectrum % Select first half

if rem(nfft,2) % nfft odd

select = (1:(nfft+1)/2)';

else

select = (1:nfft/2+1)';

end

fft_spectrum = fft_spectrum(select,:);

freq_vector = (select - 1)*Fs/nfft;

end

Answer 1

[Star Strider]

I have not run the code, however I believe the scatter call used to plot the peaks has the arguments reversed.

Try this instead —

scatter(locs,peaks,'^r');

That should work.

.