vibration time Synchronous averaging

[Faris_Elasha]

Hello

I have project to time sychronous average of gearbox signal, the signal from accelerometer is segmented every certain shaft rotation, the rotation of shaft is measured using  tachometer, then the segments average, the whole process is shown in  image below, could please provide some hints on how to apply this in labview

[Faris_Elasha]

it seem no one come accross this post, I am wondering if can use trigger for this purpose and how to to signal segmentation

Regards

[TrottM]

Hey Faris

 To make any meaningful suggestions I think we need a bit more info from you. What are you using to aquire your data? is it in DAQmx or Real-Time?

The way I would implement this however is a producer consumer type architechture. Your producer loop would be continuously aquire your vibration data and then write it to a queue. Your consumer loop would then be reading your tachometer and dump the contents of the queue into an array that could then be processed somewhere in the code.

hope that gives you a starting point.

[Faris_Elasha]

Thank TrottM for replay

to make it Clear; I use DAQmx to acquire the data. And I need to average  together a series from signal segments ( vibration signal) each corresponding to one period of synchronizing signal ( tachometer signal which indicated one rotation of the shaft completed). In another words constant no of samples will be acquired at each shaft rotation and then this samples added to previous samples from previous shaft rotations, and then divided by the total no of shaft rotation. so the problem how to acquire same no of the points at each shaft rotation and then how to do segmentation.

Regards

[TrottM]

from the sounds of it you are after something along the lines of this then.

https://decibel.ni.com/content/docs/DOC-5311

Your trigger would be the tachometer and the the vibration data would be your analog input

This would have to assume that the tacometer is at a constant velocity to ensure that the number of samples is the same however you do this though. This is beacause you are using a clock that is not tired to the tachometer for your acquisiton.

One way round this that I have seen is to use a second tachometer that produces many more pulses per revolution on the same mounting point as the first so that you have an external clock that is sychronised.

[Henrik_Volkers]

Have a look at the ADDON sound and vibration vis.Sounds like order analyses ...

If the rotational speed is constant you can use the Threshold detector vi  to get the indices of the tacho pulses , use that data to slice the vib signal and avarage them.

If the tachosignal is not constant more math is involved. What needs to be done is a resampling of your signals .

[preston_johnson]

thanks Henrik and Matt,

Matt - your triggered acquisition method works well for steady state speed machines - where the speed does not vary.  With a controlled speed machine (without speed changes) it is possible to calculate the number of samples and sample rate such that you get very close to an exact integer number of rotations and then can segment the block into rotations and average against each other. 

Henrik - you are correct, that if speed changes, even just a little bit, order analysis (Sound and Vibration Measurement Suite), http://sine.ni.com/nips/cds/view/p/lang/en/nid/203624, has a function that takes samples measured in time (what DAQmx does - 102400 samples per second for example) and re-samples them to angular position of the shaft, (360 points per revolution for example).  Then it is easy to perform time synchronous averaging on a rotation by rotation basis, even with a continious stream of data.  

Of course, bearings produce non-synchronous vibration, and TSA will remove bearing signatures.  It is good for syncrhonous signals.  However, if you perform TSA, and then subtract the TSA average from the original signal (on a rotation by rotation basis, you now have the bearing signals and noise left over.  This is useful when the gear mesh vibrations are "loud" and you also want to evaluate the bearing signatures. 

Good Luck!

Preston

[Henrik_Volkers]

Homework?? However ..

[Henrik_Volkers]

ups there was a bug in that quick shot .....

here the corrected version ...

[Faris_Elasha]

thank you all for these helps, Henrik could you please send the file in 8.5 version as I use old version