What does an FFT of an acceleration signal show?

[trn09]

I have an acceleration signal from a wii remote and I am supposed to do an FFT (Fast Fourier Transform) on it, but I don't really understand what it means I get that. I know that a Fourier Transform takes time domain data and Transforms it to Frequency domain but I don't understand what is being done to the signal.

 

[zhermes]

Your title says, 'why'; but it sounds more like you're asking 'how' is an FFT done... can you elaborate on what you don't understand please?

The 'why' is to find out frequency information about the signal (e.g. is it periodic? if so, what frequencies are dominant? etc). Frequency information, in general, is a powerful way of characterizing a signal.

 

[trn09]

I guess I don't really understand how to interrupt the results of the Fourier Transform.

I know that with the acceleration signal after taking the Fourier Transform I would get a graph of acceleration vs frequency but what does that mean?

For example if you had a signal from an accelerometer that took data on the acceleration of a car over a time interval and you took the Fourier Transform of the signal and there was a peak in the graph at 600 Hz and 50 m/s^2. How does that relate back to the acceleration that the car experienced?

 

[zhermes]

a signal from an accelerometer that took data on the acceleration of a car over a time interval and you took the Fourier Transform of the signal and there was a peak in the graph at 600 Hz and 50 m/s^2. How does that relate back to the acceleration that the car experienced?

That would tell you that the car experienced a periodic acceleration at a frequency of 600 Hz, with a magnitude of 50 m/s^2.

Lets think of something more realistic (that kind of signal would destroy the car ;).

If you saw a Fourier transform that had peaks at:
A) 120 Hz of 0.1 m/s^2
B) 0.01 Hz of 4 m/s^2
and
C) 1 Hz of 0.5 m/s^2

Then I'd suggest perhaps:
A) is due to a feature of the accelerometer, sampling 120 times per second, inducing an artificial signature.
B) might be something real, like stopping and going at stoplights, roughly every 1/100 seconds, with an average acceleration around 4 m/s^2
C) Might be something else real, like some feature of the car causing a periodic acceleration... maybe a cylinder misfiring every once in a while, or something like that.


Does that help at all?

 

[trn09]

Thank you that did help a lot actually.