Re : signal processing problem ( starting in labview )

[mehdi33]

Hi!

I work on a subject of detection of unbalance faults of a speed rotary tool holder that ranges from 1000 rev / min to 7000 rev / min. The sensors I use are : three accelerometers in X , Y, Z and two inductive sensors. The goal is to analyze the signals from the sensor by treatment of suitable signal. I have chosen as sampling frequency 25 kHz. My problem is at the spectral analysis of signals from accelerometers, I do not know if I made a mistake in my program on Labview . The spectra obtained have too much noise spectra unlike inductive sensors which give satisfactory results . attached is my work on Labview and the spectra obtained from an accelerometer and an inductive sensor. Your advice will be really useful to me !

Regards,

[Niki']

Hi mehdi33,

It seems indeed that your signal is pretty noisy. I am unfortunately not an expert in accelerometer, and I don't see anything wrong in your diagram. After few researches I found that getting noise for accelerometer is quite commun. You way have to filter it. Can you tell me what accelerometers are you using ?

Regards.

[mehdi33]

Hi Niki,

Thank you for your answer, I use a ICP acceleromoters !

Regards,

[Bob_Schor]

What You See is What You Measured (Wysiwym).  Your input is a TDMS file, so the data have already been sampled, and the spectra are whatever they are.  If you are plotting Spectra, you should note that the X Axis should be labeled "Frequency", not "Time" (to avoid confusing the reader).  

Your Spectra are a little strange, again possibly because of the units (and scaling).  What is being plotted in the "clean" spectrum (with about 5 very narrow spikes)?  Can you explain this spectrum?  For example, is the motor running at a constant speed, and if so, what is the speed?

If you have an accelerometer mounted on the motor, it will sense vibrations.  If there is any out-of-balance in the motor, it will induce vibrations consistent with the speed of the motor.  Quite frankly, I would expect an "out-of-balance" motor to show an acceleration spectrum with a peak at the rotation frequency (since it is being "shaken" once per revolution) and also to have significant high-frequency "vibration" noise.  

I'm not a "vibration" person, so I don't understand the purpose of the SVL Integration function.

Bob Schor

[mehdi33]

Hi Bob Schor,

Thank you for your analysis to my problem. Indeed the speed of rotation is variable ( 1000 rev/ min - 2000 rev/ min - 3000 rev/ min -....- 7000 rev/min) , the time for each acquistion rate is 30s . the aim of SVL Integration is to deduce the movement of the rotary tool from its acceleration by a double integration.

Regards,