Find min and max values in a constant interval.

Hello all,

i have a dataset, which records Torque in a cyclic rotation, which creates a sinus signal. It writes every 30 seconds 512 Values and then stops for 30 seconds again.

This dataset records constantly the required torque, but i only need the max and min of the sinus curve for each cycle. I could not make findpeaks() work on this dataset. Could you help me to extract max and min values of each cycle? I added a small section of the data.

Regards

[KSSV]

How did you read the file?

[Dzhem Kurtulan]

With import file option and then as column vectors.

Answer 1

[Star Strider]

The findpeaks function works, however it needs help to detect correctly the maxima and minima.

Try this —

opts = detectImportOptions('https://www.mathworks.com/matlabcentral/answers/uploaded_files/649685/Example.txt', 'DecimalSeparator',',', 'VariableNamingRule','preserve');
T1 = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/649685/Example.txt',opts)

T1 = 1536×5 table
     Sec         mm         N       N-mm     segments
    ______    ________    ______    _____    ________

    30.421    -0.21716    -19508    72081      147   
    30.423    -0.21744    -19509    72444      147   
    30.425     -0.2175    -19510    72486      147   
    30.427    -0.21666    -19509    72702      147   
    30.429    -0.21753    -19510    72646      147   
    30.431    -0.21773    -19509    72706      147   
    30.433    -0.21825    -19510    72574      147   
    30.435    -0.21739    -19508    72533      147   
    30.437    -0.21806    -19506    72399      147   
    30.438    -0.21821    -19506    72374      147   
     30.44    -0.21829    -19505    72291      147   
    30.442     -0.2181    -19505    72205      147   
    30.444    -0.21642    -19505    72104      147   
    30.446    -0.21646    -19507    71872      147   
    30.448    -0.21606    -19507    71568      147   
     30.45    -0.21704    -19509    71007      147   

t = T1.Sec;
trq = T1.('N-mm');

figure
plot(t, trq)
grid
xlabel('Time (s)')
ylabel('Torque (N-mm)')
title('Original Signal')

gaps = [find(diff(t)>0.01); size(T1,1)];                        % Time Discontinuities
segs = [[0;gaps(1:end-1)]+1 gaps]                               % Segment Indices

segs = 3×2

1 512 513 1024 1025 1536

segnrs = size(segs,1);
figure
for k = 1:size(segs,1)
    idxrng = (segs(k,1):segs(k,2));
    [pks{k},plocs{k}] = findpeaks(trq(idxrng), 'MinPeakProminence',250);            % Maximum Values & Relative Indices For Each Segment
    [vys{k},nlocs{k}] = findpeaks(-trq(idxrng), 'MinPeakProminence',250);           % Minimum Values & Relative Indices For Each Segment
    
    subplot(segnrs,1,k)
    plot(t(idxrng)-t(segs(k,1)), trq(idxrng))
    hold on
    plot(t(idxrng(plocs{k}))-t(segs(k,1)), pks{k}, '^r')
    plot(t(idxrng(nlocs{k}))-t(segs(k,1)), -vys{k}, 'vr')
    hold off
    grid
    title(sprintf('Segment #%d',k))
    ylabel('Torque (N-mm)')
    text(t(idxrng(plocs{k}))-t(segs(k,1)),  pks{k}, compose('  \\leftarrow %+6.0f',pks{k}), 'Horiz','left')
    text(t(idxrng(nlocs{k}))-t(segs(k,1)), -vys{k}, compose('  \\leftarrow %+6.0f',-vys{k}), 'Horiz','left')
end
xlabel('Time (s)')

The code is reasonably straightforward. The only complications are the indexing for the various segments in the loop. The maxima and minima are in the ‘pks’ and ‘vys’ cell arrays respectively, and are displayed on the plots.

.

[Dzhem Kurtulan]

Hello Star Strider, thank you for your answer.

I uploaded a part of my data, but actually data is much bigger, around 700MB.

What i am trying to do is, to plot the drawn lines on the original data. So plot of max and mins. Your cell array pks and vys has 3 values for each column, so i am a bit confused about it, could you explain me that?

[Star Strider]

I interpreted the information in the original post as as best as I could. My code provides the maxima and minima for each cycle, as requested, and demonstrates how to retrieve the results of that analysis.

To calculate and plot the maxima and minima of the entire data set (and not of the individual cycles), use the envelope function.

Answer 2

[KSSV]

If A is your complete matrix.

t = A(:,1) ;   % this is seconds column
iwant = A(t==30,:) ;   % extract chunk which is for 30 secs

Now you have data of 30 seconds in hand, you can find max and min of the column you want.