Peak RMS or RMS or absolute average

[Luppy]

Hello guys

I'm doing some signal processing/data analysis. Hope to get some advice on something.

I collected some vibration data just the other day while metal cutting and I'm wondering what is the correct (or better) way to do this.

These are what the signals look like. Sampling rate of the data is 10000 hz.

Upclose,
https://i.imgur.com/93g8L.jpg
The signals kinda repeat themselves every 0.0377 seconds. That's the time taken for 1 revolution of the cutter.

If I'm trying to compare the changes in the general vibration level when metal cutting parameters are varied (When plotted, vibration on the Y axis against variation of machining parameter on the X axis) , would it be better to use the
1) the rms of all the data points for a fixed number of revolutions of the cutter
2) or the rms of just the peak values for a fixed number of revolution of the cutter
3) absolute average of the peak values for a fixed number of revolution of the cutter

All 4 methods would give me a different value.

This probably doesn't matter that much as long as I keep the method constant. But, it still got me thinking for a few days. Any help greatly appreciated

 

[FvM]

Obvious question: Which information behind the data you are looking for?

You probably would want to look for a spectral representation of data, like this product does.
https://www.qass.net/cms/index.php/en/products/optimizer-4d

 

[Luppy]

I'm just trying to see a variation of general vibration level when I change machining parameters like the depth of cut.

 

[rohitkhanna]

Obvious question: Which information behind the data you are looking for?

You probably would want to look for a spectral representation of data, like this product does.
https://www.qass.net/cms/index.php/en/products/optimizer-4d

I agree. A spectral analysis which shows independent frequency/ amplitude components will likely give you the best insights on your vibrations.
You could do a static analysis on your collected data, or a continous one like that referred by FvM.

 

[FvM]

If you are not looking for specific information, I agree with your previous assumption:

All 4 methods would give me a different value.
This probably doesn't matter that much as long as I keep the method constant.

I'm just trying to see a variation of general vibration level when I change machining parameters like the depth of cut.

Of course you'll see a variation. But what does it mean? That's where time resolved spectral analysis (in combination with a theoretical analysis of the metal cutting process) comes into play.