how to double integrate highly nonlinear numerical dataset

Dear all,

I need to find displacement data from the available acceleration dataset. All dataset is numerical and highly nonlinear.

I was wondering if there any efficient methods with the least error (e.g., trapezoid method may generate too errorneous answers.).

AND more importantly, are there any features in the signal processing and communucation app in matlab that facilitates such a signal processing to cinvert the acc. data into disp. data?

Thanks!

[Torsten]

And I guess you have very noisy data for acceleration ?

[John D'Errico]

If your data is noisy, then you DON'T want to use a higher order integration. In fact, a trapezoidal rule may be best. Yes, there is integration error, but if you are integrating noise, then you are still getting garbage. And the double integral of garbage is garbage piled deeper and higher.

[Mathieu NOE]

hello @Alireza Babaei

most of the issue is to handle properly the DC signal (in your acceleration signal what is "real" and what is offset error or drift , and how to avoid integrating them). without having an idea of what the result should be (in others terms, a tool that helps you calibrate your post processing) it's almost mission impossible to get this right.

using acceleration record, Get velocity and displacement signal & carry out FFT and plot the spectra - MATLAB Answers - MATLAB Central (mathworks.com)

Issue with magnitude of cumtrapz and integration of second order system acceleration data - MATLAB Answers - MATLAB Central (mathworks.com)

[Alireza Babaei]

Thanks all, it was good hepl.

Yes the collected data is still noisy though we have utilized very expensive accelerometers.

I think as you mentioned there is no feature on the signal processing app in matlab and I have script the proper code (and I will utilize the ones you sent to me) to handle this properly.

Thanks a lot all!