I have "step" data and I want value only on the steps

I have a data (that came from a misure of a displacement) and I need to extract the value of the data of every step. I tried to use "findpeaks" but it can't detecte every step (like you can see from the figure below) ((ignore Data fitting)).

The step should be 45 because there is a motor that change is position according to the angle.

In the file "DATI" there is the matrix from where the data came. First column is the displacement, second colomun is the time of each recording data (third and fourth column are irrilevant to the problem but they are the angles and respective times, as you can see, the two times aren't syncronized).

I tried to use this code but can't get any better:

mis = importdata("excel\DATI.csv");

x_mis = mis(:,1);

x_t_x = mis(:,2);

% The value that I need end at the 8022th row

x_mis_n = x_mis(1:8022,1);

x_t_x = x_t_x(1:8022,1);

% Trying to search for "plateau" value of each step

[pks,locs] = findpeaks(x_mis_n,x_t_x,'MinPeakDistance',0.3,'MinPeakWidth',0.01);

Answer 1

[Star Strider]

First, detrend the curve, second, find the peaks, last, return the values.

Try this —

mis = importdata("DATI.csv");
x_mis = mis(:,1);
x_t_x = mis(:,2);

% TF1 = islocalmax(x_mis, 'FlatSelection', 'first');
% 
% figure
% plot(x_t_x, x_mis)
% hold on
% plot(x_t_x(TF1), x_mis(TF1), 'rs')
% hold off
% grid

% The value that I need end at the 8022th row
x_mis_n = x_mis(1:8022,1);
x_t_x = x_t_x(1:8022,1);

x_mis_n_dt = detrend(x_mis_n, 2);
TF2 = islocalmax(x_mis_n_dt, 'MinProminence',0.05);               % Use: 'islocalmax'
[pks,locs] = findpeaks(x_mis_n_dt, 'MinPeakProminence',0.05);       % Use: 'findpeaks'

figure
plot(x_t_x, x_mis_n)
hold on
plot(x_t_x(TF2), x_mis_n(TF2), 'rs')
hold off
grid
xlabel('x\_t\_x')
ylabel('x\_mis\_n')
title('Results: ‘islocalmax’')

Peak_Values = table(x_t_x(TF2), x_mis_n(TF2), 'VariableNames',{'x_t_x','x_mis_n'})

Peak_Values = 30×2 table
    x_t_x     x_mis_n
    ______    _______

    4.0622    0.1995 
    5.3659      0.35 
    6.0477     0.532 
    6.7004    0.6615 
    7.3363    0.8995 
    8.0061     1.071 
    8.6254     1.267 
    9.2743     1.463 
    9.9239      1.82 
    10.568     2.079 
    11.214    2.3415 
    11.863     2.625 
    12.516     3.024 
    13.252     3.185 
    13.841    3.5315 
    14.497     4.032 

figure
plot(x_t_x, x_mis_n)
hold on
plot(x_t_x(locs), x_mis_n(locs), 'rs')
hold off
grid
xlabel('x\_t\_x')
ylabel('x\_mis\_n')
title('Results: ‘findpeaks’')

Peak_Values = table(x_t_x(locs), x_mis_n(locs), 'VariableNames',{'x_t_x','x_mis_n'})

Peak_Values = 30×2 table
    x_t_x     x_mis_n
    ______    _______

    4.0622    0.1995 
    5.3659      0.35 
    6.0477     0.532 
    6.7004    0.6615 
    7.3363    0.8995 
    8.0061     1.071 
    8.6254     1.267 
    9.2743     1.463 
    9.9239      1.82 
    10.568     2.079 
    11.214    2.3415 
    11.863     2.625 
    12.516     3.024 
    13.252     3.185 
    13.841    3.5315 
    14.497     4.032 

% figure
% plot(x_t_x, x_mis_n)
% hold on
% plot(x_t_x(TF2), x_mis_n(TF2), 'rs')
% hold off
% grid
% axis([15 20 3 7])

I initially tried islocalmax with the 'FlatSelection','first' name-value pair, however it did not find all of them. My detrending approach seems to work.

.

[Farotto Luca]

Thank you very much, it works like I was thinking!

[Star Strider]

As always, my pleasure!

Answer 2

[the cyclist]

I am not sure how you want to define the "plateau" value, since the displacement moves both up and down. Can you just use the unique values of the displacement?

dati = readtable("DATI.csv");

time = dati.Var2;
displacement = dati.Var1;

plot(time,displacement)

% Plateaus defined as unique displacement values
plateau = unique(displacement)

plateau = 1295×1

-0.0595 -0.0525 -0.0490 -0.0385 -0.0350 -0.0280 -0.0245 -0.0105 0 0.0035

The graph of all the data looks different from yours (including having some negative values for displacement). I don't know how you want to manage that.

[Farotto Luca]

Sorry that I didn't write it before and specified it only in my code, the values that I need are only to the 8022 row, so it only mooves up .

For "plateau" I mean every time that the value has a costant value.

I tried to use unique displacement values but it takes out too much values in compare to the steps

[Sam Chak]

@Farotto Luca, Do you mean these 'plateau' values?

[Farotto Luca]

Yes, I meant them!

Answer 3

[Aaron]

You might want to try locating the peaks on the difference of the data rather than the data itself.

I'm not familiar with the findpeaks function, but here is a small example of locating data at steps with diff:

(Note that whether you insert the NaN before or after the diff(y) affects whether you're looking at the data before or after the jump discontinuity.)

x = 1 : 100;
y = reshape((1 : 10).*ones(10,1), 1, []);   % Step data
dy = [NaN,diff(y)];
peaks = dy > 0.5;                           % Locate jumps greater than 0.5

figure; hold on
plot(x,y)
plot(x(peaks),y(peaks),'or')

Answer 4

[Austin M. Weber]

If you want to count each step as being any increase from the previous value then this is what you want:

locs= [];
for i = 2:length(x_t_x)
	if (x_mis_n(i) - x_mis_n(i-1)) > 0
		% Identify position of step
		locs = [locs i];
	end
end
pks = x_mis_n(locs);

Or, without the loops:

d = diff(x_mis_n);
locs = find(d > 0)+1;
pks = x_mis_n(locs);