acceleration versus frequency question

Oct 12, 2003 3 Replies

Hi,



I have a time series of accelerations (acceleration versus time) and I need to convert it to acceleration versus frequency. At a glance my first idea was to take the Fourier transform of the data set which would give power versus frequency and then use some kind of existing proportionality to get acceleration versus frequency. The first step (taking the Fourier transform) is trivial but I'm not sure how to proceed after that. Furthermore, I'm not even sure if this approach is correct. Could someone who has done something similar to this in that past give me an idea how to proceed? Thanks.



Rory



You are on the right track. Getting the calibration right is tricky. One way to look at it is to use energy compensation. The total energy in the spectrum should be the same as the total (RMS) energy in the time history.

This means that the height of the peaks will be 'wrong', but you can't have everything.

Incidentally, when you are trying out a new FFT always analyse a pure sine wave to see if you get what you are expecting.

Cheers

Greg Locock

What am I missing? Why is the transform of a(t) => P(w)?

jim andrews

Nothing, I think.

Um, that's an unanswerable question. try alt.sci.why.is.the.universe.the.way.it.is

You could try the links in

formatting link
More helpfully, bung me a copy of your data in some sort of ascii format and some details of calibrations and I'll try and give you a result, and explain how I got there. I have a feeling you might be worried about philosophical details that we tend to run roughshod over.

Cheers

Greg Locock

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required