Peak to peak amplitude

I have registered an oscillating angle and I need to calculate the range of movement of such angle. The signal looks like this:

I would like to calculate individually the amplitude between a peak and the following valley, I have tried several functions and transforms (hilbert, fourier...) but I am not sure if it is the ideal solution mathematically speaking. I am looking for a simpler, point-to-point calculation. Also, some peaks are "twin peaks" and I would like to ignore such events without filtering the data. I have attached a sample part of my data.

Any ideas,

Thanks in advance.

Answer 1

[Star Strider]

LD = load(websave('question_sample','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1270450/question_sample.mat'));
Data = LD.MATLAB_question

Data = 17558×2

1.0e+06 * 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000 1.0226 0.0000

x = Data(:,1);
y = Data(:,2);

[pks,plocs] = findpeaks(y, 'MinPeakProminence',8, 'MinPeakDistance',100);
[vys,vlocs] = findpeaks(-y, 'MinPeakProminence',7, 'MinPeakDistance',100);

vlocs = vlocs(vlocs > plocs(1));
P_P = pks - vys;

pvlocs = round(mean([vlocs plocs],2));

figure
plot(x, y, 'DisplayName','Data')
hold on
plot(x(plocs), pks, '^r', 'DisplayName','Peaks')
plot(x(vlocs), -vys, 'vr', 'DisplayName','Valleys')
plot(x(pvlocs), P_P, '.-m', 'DisplayName','P-P')
hold off
grid
legend('Location','best')
xlim([min(x) max(x)])

PksVysPP = table(x(plocs), pks, x(vlocs), vys, P_P, 'VariableNames',{'X Peak Location','Peaks','X Valley Location','Valleys','P-P'})

PksVysPP = 36×5 table
    X Peak Location    Peaks     X Valley Location    Valleys      P-P  
    _______________    ______    _________________    ________    ______

      1.0228e+06       10.696       1.0229e+06         0.82368    9.8726
      1.0232e+06       12.849       1.0234e+06          3.1877    9.6609
      1.0237e+06       8.5264       1.0238e+06           2.923    5.6033
       1.024e+06       16.024       1.0243e+06        -0.25246    16.276
      1.0245e+06       16.536       1.0247e+06         -1.8931    18.429
      1.0249e+06       23.716        1.025e+06         -5.1215    28.837
      1.0254e+06       19.499       1.0257e+06         -1.8049    21.304
      1.0261e+06       17.471       1.0263e+06         -2.7576    20.228
      1.0268e+06       17.206        1.027e+06        -0.25246    17.458
      1.0275e+06       15.248       1.0276e+06         -3.9748    19.223
      1.0281e+06       15.971       1.0283e+06          -2.246    18.217
      1.0288e+06       15.424       1.0289e+06         -3.3574    18.782
      1.0295e+06       14.613       1.0296e+06         -2.9516    17.564
      1.0301e+06       18.123       1.0303e+06          -1.311    19.434
      1.0308e+06       12.796       1.0309e+06          1.4941    11.302
       1.031e+06       10.749       1.0314e+06          2.3056    8.4436

This may be a bit fragile, since the findpeaks calls require two name-value pair arguments to get the desired results. It appears to work here.

.

[Marc Elmeua]

Thanks a lot Star Strider. This is very close to what I am looking for, and it seems to work just right for the moment. The only thing is that the valleys appear to be negative in the table, I have corrected this by writing

vys = -vys

after the findpeaks function. I am not sure if this is very clean but it seems to work.

Again, thanks for the help.

[Star Strider]

As always, my pleasure!

I used the signs of ‘vys’ correctly in the plot and calculations except in the table. My apologies for that overisght. (The table was an afterthought.)

With noisy data that is not filtered, it is necessary to make certain adaptations. For that reason, the code is a bit fragile, so unless other data sets have essentially the same characteristics, the name-value pair argument values may need to be tweaked.

.