How to remove a certain value from all entries of a vector

Hi,

I have given different vectors by measuring mechanical stress in an experiment.

However, due to a measurement error, an offset occurred after a few seconds (see following picture):

Right now the code looks something like this:

x60l = Schnitt_60_1.Schnitt_60_1.Zeit;

x60 = x60l(1:end);

a60l = Schnitt_60_1.Schnitt_60_1.DMS_1;

a60 = a60l(1:end);

B2 = smoothdata(a60,"movmean","SmoothingFactor",0.2,"SamplePoints",x60);

plot(x60,B2,'Color','[1 0.5 0]','LineWidth',1.5)

Now I would like to delete the first values of my vector and adjust the remaining values so that the signal starts at 0 again (removing the offset). How can I do this?

Thank you in advance

Answer 1

[Cris LaPierre]

You might consider using the Remove Trends live task if you are working in a live script. This is limited to removing polynomial trends. The live task makes it easy to test different settings quickly to try and find one that works for you. You can see an example here.

[Elias Klee]

Elias Klee weniger als eine Minute ago

Hi,

I am really just a beginner when it comes to MatLab. That's why I was hoping for a little more intentional approach, such as subtracting a specific scalar of every value of the vector a60 in order to remove the offset.

I don't really feel like removing trends is what I am trying to do.

Do you have any ideas how to remove the offset (which is not a trend in my understanding) in a little more vector-oriented way?

Thanks a lot anyways for your answer

[Cris LaPierre]

Just subtract a scalar value from all your data.

a60 = a60l(1:end) - 30;

You get to pick what the scalar value is.

Answer 2

[Star Strider]

You can remove the offset and the trend with a highpass (or bandpass) filter. (Use bandpass if you want to elimiinate high-frequency noise as well.) For best results, use the 'ImpulseResponse','iir' name-value pair.

First use the fft or pspectrum function to understand the frequency content of the signal. This will allow you to determinne the filter stopbands correctly. It will likely be necessary to experiment to get the desired result.

All of these (except fft) require the Signal Processing Toolbox.

[Elias Klee]

Hi,

I am really just a beginner when it comes to MatLab. That's why I was hoping for a little more intentional approach, such as subtracting a specific scalar of every value of the vector a60 in order to remove the offset.

I don't really have the need to filter out specific frequencies nor to remove a trend. Do you have any ideas how to do it in a little more vector-oriented way?

Thanks a lot anyways for your answer

[Star Strider]

To subtract a specific value, MATLAB implicitly expands the operations automatically so it can be done in one operation —

x = 0:25;                                   % Create Data
y1 = x+5 - x/100 + randn(size(x));          % Create Data

figure
plot(x, y1)
grid
ylim([-30 30])

y2 = y1 - 15;                               % Subtract Constant

figure
plot(x, y2)
grid
ylim([-30 30])

.