How can I obtain the difference between two data stored in workspace?

I experimentally tested a motor and the measured speed and the reference speed were stored in the workspace. My supervisor asked me to obtain the speed error (the difference between the reference and the measured speed). How can I obtain it from the data stored in the workspace?

[saleh shlimet]

I could obtain the difference speed and separately plot two errors (error and error1) for two different methods, but how can I plot these two errors togather on the same time axes? see attached

[Star Strider]

I do not understand the problem.

Using the hold function could be one possible solution.

[saleh shlimet]

error=novel speed-reference speed, error1=classical speed-reference speed. to plot error we use this command "plot(error)". I want to plot error and error1 together? I tried this command "plot(error,error1)", but it did not work.

[saleh shlimet]

thanks I tried hold on and it is working.

[Star Strider]

As always, my pleasure!

Answer 1

[Star Strider]

If the two variables have the same number of elements, just subtract one from the other.

For example:

speed_error = reference - measured

That will give the error at each element of each vector.

If you want to get a single parameter with respect to the ‘speed_error’ vector, one option is to use the rms function to get the RMS error. To get the mean squared error, simply square the RMS result.

There are other options as well. The sum squared error is simply:

sum_squared_error = sum(speed_error.^2)

.

[saleh shlimet]

(If the two variables have the same number of elements). Do you mean the time scale?

[Stephen23]

"Do you mean the time scale? "

As Star Strider explained, in general to subtract (i.e. get the difference between) two arrays then they should have the same number of elements. I would go even further and suggest:

isequal(size(reference),size(measured))

[Star Strider]

@saleh shlimet — If they are sampled at the same times, or if the sampling frequecies are regular (all the sampling intervals are the same in each vector) and the sampling frequencies are the same between the two vectors, yes, that as well.

@Stephen23 — Good point!

[saleh shlimet]

If the two variables do not have the same number of elements? how to find the error?

[Star Strider]

Since they both by definition have to have the same sampling frequencies (so the differences between their respective time data are constant), the only way to calculate the error in that instance is to shorten the longer vector to the length of the shorter vector. Interpolating one ot the length of the other will not be appropriate.