STFT in rotordynamics

[heitor]

Hello.

I've heard that STFT is a very useful tool to analyse the dynamics of a rotor. Is that true? If yes, can you recommend me some bibliography?

Thanks!

 

[electricpete]

What does STFT stand for?

 

[GregLocock]

Oh electricpete you spoiled my game. I was going to let it go unanswered till he figured that part of the problem out for himself.

heitor - who did you hear it from? why not ask them? do they often lie to you?



Cheers

Greg Locock

 

[heitor]

You have a very nice way to say 'I don't know'.
STFT means short time fourier transform, and when I find something about using it on rotordynamics, I will post it here .

 

[anands78]

http://engineering.rowan.edu/~polikar/WAVELETS/WTtutorial.html

Has some material on STFT..I cant think of any application to RD.

 

[GregLocock]

heitor - OK I was being funny. I've been doing rotating machinery analysis for 20 years, so if I haven't seen an abbreviation I think it is fair to say it is a bit specialised.

Anyway, yes we use different resolutions for our FFTs, but frankly the use of very short frames of data is rarely used in the automotive world since we are often very interested in the exact frequencies, and our slew rates are not very high. If your slew rates are high enough that you need to use STFT then I suggest you probably need to think about a time domain analysis not an FFT. You should probably be thinking about using a gated sampling technique at the very least (the diesel guys use this to look at the combustion noise signature from each cylinder).

Incidentally anybody who hasn't read polikar's web page is in for a treat, it is the best discussion of ffts and wavelets I've seen.

Cheers

Greg Locock

 

[kmgary]

I have been doing some research on non-stationary vibration analysis.STFT is one of useful time-frequency analysis tools to get the information of rotordynamic during run-up and coast down.

 

[heitor]

Thank you, GregLocock

 

[Tunalover]

Greg Locock-
Can you provide Polikar's web page address with its discussion of ffts and wavelets?

Tunalover

 

[GregLocock]

It's the one given in anands78 post. I've been usimg that website as a reference for guys at work for a couple of years, in fact just about every time wavelets come up. It has a great discussion of why the inherent limits that apply to ffts also apply to wavelet analysis.

Cheers

Greg Locock