Continuous wavelet transform of vibrational signals (scalogram)

Hello to everyone,
I am a beginner in signal processing and I'm trying to perform a CWT of a vibrational signal, took on a IC engine, in order to better understand the differences existing with the time-frequency rappresentation. I performed the scalogram, but i do not understand why on the time axis i read wrong informations, that is, there are wrong values on the time axis. Is there anyone that can help me?
Thanks to all.
Attachments:
Scalogram.png ‏51 KB
Scalogram VI's front pannel.png ‏184 KB

Thanks for the answer,
I mean that i don't see the time values of the signal's time history.
Are they the traslation factors used to perform the CWT?

Similar Messages

How to use wavelet transform in vibrational signals?

plzzz....show the explaination of wavelet transformation in vibrational signals and also help me in understanding the waveforms......................

I am not a wavelet expert, yet I would experiement with the example in the application LLB for engine knock. In general wavelets are used to extract impact events in vibration signals.
Do you have a data file of your vibration data? What are you hoping to detech?
Preston Johnson
Principal Sales Engineer
Condition Monitoring Systems
Vibration Analyst III - www.vibinst.org, www.mobiusinstitute.com
National Instruments
[email protected]
www.ni.com/mcm
www.ni.com/soundandvibration
www.ni.com/biganalogdata
512-683-5444

Continuous Wavelet Transform

English: I have the Signal Processing Toolset as well as LAb View 7.1.I want to use Continuous Wavelet Transform to get a time-frequency representation of the signal (using Morlet wavelet for example), just like with the spectrogram. How can I do this? Are there any demo-files?
German: Ich habe das Signal Processing Toolset und Lab View 7.1. Ich möchte die Continuos Wavelet Transform nutzen um einfach eine Darstellung der Amplitude über Zeit und Frequenz zu erhalten (z.B. mit Morlet Wavelet), so wie beim Spektrogramm. Frage: Wie realisiere ich das? Gibt es vielleicht ein Demo-File?

I do not have the SPT 5 at hand. I just copy the diagram and front panel of CWT in SPT 7 for your reference.
Attachments:
CWT Example.PNG ‏125 KB

Continuous Wavelet Transformation frequency output in columns and rows?

Hi
I´m trying to understand the output of the Continuous Wavelet Transformation Vi. I am only a beginner in wavelet, so is viewing from a Fourier Transformation method with moving windows. For example if I have 8 points say recorded with 1Hz = 8sec of data and run it thought the CWT with 8 scales, I get 8x8 array of data. So as far as I understand I get the frequencies of (starting with column 0 going to 7): 1Hz, 0,5Hz, 1/3Hz, 1/4Hz, 1/5Hz, 1/6Hz, 1/7Hz and 1/8Hz. Is this correct so far?
What I don´t get is how it is possible to make 8 (rows) of wavelet correlations on the columns from 1-7, when there is for example only one possibility to make correlation of 1/8Hz stretched wavelet, because the wavelet must be stretched over 8points (7th column). In picture below is illustration of wavelet where for the largest scale 8 (=lowest frequency) here 1/8Hz there is only 1 value for every 8 values for the smallest scale 1 (=highest frequency) here 1Hz.
I really hope someone can help me!
Best Regards
Jesper

Your current code can be replaced by an autoindexing FOR loop, you only need once instance of the "add array elements" operation. Easiest would be to acquire it as a 1D array with a constant number of samples then use "reshape array" to create a 2D array of the desired dimensions to be used for autoindexing as suggested.
LabVIEW Champion . Do more with less code and in less time .

Question on the result of using Wavelet transform on sine wave

Dear all,
I have apply the Wavelet transform on a 50Hz sine wave. The result is shown below. But I don't understand the result of Wavelet transform. Anyone can help me? Thanks.
Victor

You seem to have something against the FFT, yet continually compare other transforms to the "sweet spot" for the FFT, ie. sinusoidal and periodic functions. Without trying to write a treatise on the subject (there are many fine books), a few observations.
The primary purpose of any transform pair is to perform a rotation in function space, for the FFT we are familiar with the time domain and frequency domain. The wavelet domain is not so simple. The FFT is very effective for sinusoidal and periodic functions, ie. those that have no localization in the time domain. If you have a sinusoidal input, the infinite signal in the time domain can be reduced to a pair of numbers (one if you take the power spectrum). That is very efficient, other periodic signals can usually be approximated by a small number of terms.
Wavelets on the other hand are chosen to be localized in both the time domain and the wavelet domain. If you look at a periodic signal like your sine wave, you see a mess, and to recreate the input signal you will need to keep an awful lot of the points around for the inverse transform. You may be able to set the last 1/3 to zero, but that's it. To see an effective application, you should be looking at localized functions, like an impulse. Put an impulse into the wavelet transform and you get something that is still a bit complex, but pretty simple. The FFT of an impulse contains components at all frequencies and the wavelet clearly wins in this case.
On paper, there is usually a clear choice, and the FFT almost always wins for spectral estimation. On the other hand, if you are trying to compress "real world" data, for instance an image, then there is a choice to be made. Standard JPEGs are FFT based, you take the FFT and try to keep just the largest frequency components. As you probably notice, edge contrast usually suffers, especially at higher compression. With wavelets, it is typical to have slightly better contrast leading to better preservation of details for a given compression.
In short, my opinion is that transforms are useful when they simplify the problem. Wavelets do not simplify the representation of a sine wave.

Normalized Frequency Analytic wavelet transform

Hi,
I am new to Labview and Signal Processing. For my work (file attached), I want to see the frequency component in detail, by using AWT Scalogram method available in signal processing toolkit wavelet transform (file attached). Signal data is read from text file and I get the frequency distribution on the scalogram. What I want is to normalize the frequency such that 50 Hz is eqaul to 1 on the y-axis of scalogram. So that total 0-1000 Hz on the y-axis shown in 0-20 values. According to theory available the normalized frequency shows scale from 0-1 but I don`t know how to do so, Need guidline in this regard.
Urgent help will be appreciated.
Attachments:
AWT.vi ‏25 KB

Hi,
I have a similar problem. I have my scalogram from an analytic wavelet transformation. Now I wanna know the min. and max. frequency.
How do i do that?
If i take the VI "Scalogram_Y-axis.Maximum", labview gives me always the max of my y-axis....
Hope someone can help me...
I added my VI, and the file from which the data comes...
Attachments:
Wavelet_Test1.vi ‏28 KB
05_07_12_Tiefpass.zip ‏31 KB

Set Coefficients of Discrete Wavelet Transform VI

Regarding the Advanced Signal processing toolbox I have a query regarding the one of the examples that can be found in the Wavelet Analysis/Getting Started exaples folder.
In this exaple a DWT is performed on a Doppler signal using the Discrete wavelet Transform VI. Secondly the Get Coefficient of Discrete Wavelet Transform VI is used to display the coefficient at the selected level, Then, and this part confuses me, the WA Set Coefficients of Discrete Wavelet Transform is used here to set the coefficients at the selected level to 0?? and why is this block diagram using a Subtract function with both of its inputs tied to the selected DWT coeff output of the Get coefficients VI, instead of using an empty awrray?
My question is why are the coefficients at the selected level set to 0 before calculating the inverse discrete wavelet transform?

Dear jquinno,
have you found any answer to your questions in the last year, please? I bumped into the same "model" questions lately, and I was wondering.
Cheers,

Wavelet transform scale and time information

Hi there -
I am using the wavelet transform for a non-stationary signal. I am not having trouble getting the coefficients from the transfrom but is there I way I can find the scale/frequency and time information from just the wavelet VI. I see no way I can get at this information. Also, I am pretty certain Labview uses the DWT when computing the transform but does Labview have the option to do the CWT?
The reason is because I want to plot time vs. frequency vs. amplitude and I need all three to do it properly.
Thanks for your help,
Cameron

Hi, Cameron.
This screenshot shows one DWT and one CWT VI, and you'll notice that the CWT has an output called scale info which contains the time information and the scale (frequency) information. (In addition, LabVIEW has several other wavelet VIs.)
If this doesn't answer your question, please let me know. Have a nice afternoon!
Message Edited by sarahk on 08-16-2006 03:59 PM
Sarah K.
Search PME
National Instruments
Attachments:
scale info.JPG ‏37 KB

How to do FFT of vibrations signal?

Hello,
I'm trying to measure the time and frequency spectrum of a vibrations signal.
My time plot seems to be ok, but my FFT graph doesn't look right.
An expected vibration FFT graph of a bearing should look like peaks at various frequencies, but my FFT graph looks only like a linear line.
Does anyone know what's wrong with my attached program?
Thank you.
Y.Yang
Attachments:
3 axis white and green sensors v2.vi ‏122 KB

Your code looks fine. I double checked how you're acquiring data, and it seems fine. I'm pretty sure that you are acquiring an FFT and the graph is just not scaled correctly. Do the following:
Right Click Waveform Graph 2 >> X Scale >> AutoScale X
Right Click Waveform Graph 2 >> Y Scale >> AutoScale Y
I've also attached an example of what I've done to get the FFT from a simulated signal.
Che T.
Applications Engineer
National Instruments
Attachments:
FFT.vi ‏67 KB
fft.png ‏43 KB

Would like to use Wavelet Transform functions

Hello:
As the title says, does Measurement Studio has wavelet transform functions. I currently have version 6.0 (full) and version 7.0 (have not used it since I am still on VB6). Thanks.
Raj

Hi Raj,
There isn't any particular "Wavelet Transform" method available in
Measurement Studio at this time. For a complete list of analysis
functions for the various versions of Measurement Studio, check out the
following link:
http://www.ni.com/analysis/cwtools_analysis.htm
Thanks Raj, have a good one!
Dan Weiland
Applications Engineer
National Instruments
www.ni.com/support
Dan Weiland

Sound and vibration signal simulator box

When i am giving 0 to 5volt output from sbRIo to the sound and vibration signal simulator box to control the speed
the fan rotates very slow but when i a potentiometer present on board(manual control) there is linear relation with the input and output...
Max RPM being 6000
when i give output from the controller previously it was running on 4000 after few days 3000rpm and then 2400 now its very low and almost 1000 RPM
may i know the solution to rectify this...

If is the fan controlled by output voltage and you have corrected value of voltage, there is no problem in code but problem is with fan.
Simyfren

Acquiring Frequency Content of Vibration Signal

Hi.
I am trying to find the frequency content of an incoming signal from a geophone which has a 500 Hz vibrational signal on it. However, using two different methods, I find that I am only given a large DC component for the frequency spectrum of the signal. I have two different methods set up in a case statement to make it easy to try each one without having to rewrite code. The first method is to just use the FFT function (in the attached picture I have a high-pass filter on the front to try and block the DC signal, but this only blocked the DC signal and did not make the 500 Hz signal any more noticeable). The second method uses the Power Spectrum function with my array data converted to waveform for input.
The first picture shows the two different methods I tried. The 1st picture with graphs is the result I get using the FFT method. The 2nd picture with graphs is the result I get using the Power Spectrum method. Any information or advice would be greatly appreciated. I can provide further information upon request.
Thanks.
Attachments:
2methods.jpg ‏41 KB
method1.jpg ‏106 KB
method2.jpg ‏115 KB

Well, first thing, it would be a lot easier for someone to help you if you were to attach your VI to a post than to send those little pictures. Take out anything that is irrelevant or proprietary, fine. Please set up labels and scales for your axes so they reflect your data, don't just use the default labels and arbitrary scales. Taking a wag that your bottom "time" axis is scaled to the number of data points (0-9999) and your signal is actually a 500 Hz sine, you are actually only taking about 200 points per wave, instead of 600, like your VI picturette says. Even though you would certainly be above the Nyquist frequency for sampling, it throws everything else into doubt because you haven't described it accurately.
Second, if I were to also assume that the tunnel into your case structure is the array you want to get frequency from (here's where attaching the working VI comes in), you are feeding one point at a time into the FT or PS functions. With one point at a time, it'll be real hard to get any more than a DC frequency. You need to feed the whole array in for the spectral analysis functions, or at least a larger portion of it. You've just got a single point, followed by 2999 zeros, at least for the FFT portion.
Cameron
To err is human, but to really foul it up requires a computer.
The optimist believes we are in the best of all possible worlds - the pessimist fears this is true.
Profanity is the one language all programmers know best.
An expert is someone who has made all the possible mistakes.
To learn something about LabVIEW at no extra cost, work the online LabVIEW tutorial(s):
LabVIEW Unit 1 - Getting Started
Learn to Use LabVIEW with MyDAQ

How can i measure the speed of the fan in sound and vibration signal simulator box

how can i measure the speed of the fan in sound and vibration signal simulator box
provided the values tach out:2 pulses/revolution and maximum speed = 6000rpm

Hi. I tried your suggestion and I could not get it to work well. In
fact it worked better without the running average (which suprised me).
I also played with the number of data points but still could not get it
to work well.
Do you have any other suggestions on how I could sort this problem out?
I
have attached a screen shot of the signal. To recap, I'm trying to
measure the'y' position of the flat regions of the signal. The signal
does not have a constant level, so this 'y' position varies with time
in a non-unform manner.
I am currently differentiating
the signal and selecting the y position when the differential is equal
to zero (within a tolerance band).
In the screenshot, the
signal has been zeroed using the method I have described above.
However, it occasionally has glitches which may be caused when the code
picks up a zero gradient at a peak or valley in the signal.
I therefore need some kind of algorithm that only measures the position of the flat regions.
Many thanks
John
Attachments:
screenshot.gif ‏121 KB

FFT Frequency Resolution vs. Graph Update Rate of a Vibration Signal

Hello,
First, what I know:
The frequency resolution (Hz) of an fft is 1/acquisitionTime (e.g. if 2 seconds of data are acquired, the frequency resolution is 1/2s = 0.5Hz).
What I'm Wondering:
Let's say I'm displaying an acquired signal on a graph and freq res = 0.5Hz. The graph update rate is therefore approx 2sec.  This may appear to be an application with a problem from a user's perspective.
Can I increase the update rate?
It seems that with vibration data (i.e. mostly non-periodic) I could create a buffer that is 2sec long.  The first data acquisition sequence will fil the buffer with 2sec of data. Then I could just acquire something shorter than 2sec of data (let's say 1/4sec). If I am acquiring data continuously and I push the 1/4sec of data into the front of my buffer then perform the fft again, I have effectively increased my graph update rate to 1/4sec and my users think I'm great.
Can anyone shed any light on whether or not (why/why not) this will work?
Thanks for your time,
Chris

The best tool NI provides for your application is the Zoom FFT. This allows you to specify the span (frequency range of the FFT) and the zoom (the delta f such as 0.5hz) in the FFT. The Zoom FFT has a built in buffer, called overlap and allows you to set the overlap so that it updates at the appropriate rate, event though you may need many seconds of data to get the delta-f in the FFT you are looking for. We package all this and manage units to display the correct units for power density, power, linear, magnitude, rms, 0 to peak, peak to peak, etc. We even provide a express VI for the zoom FFT and it is also part of the Sound and Vibration Assistant, requiring no programming at all.
http://digital.ni.com/public.nsf/allkb/F70DB3441DC5AB9986256E97005BA825
http://zone.ni.com/devzone/cda/tut/p/id/6107
http://zone.ni.com/devzone/cda/epd/p/id/5505
http://sine.ni.com/nips/cds/view/p/lang/en/nid/12152
Preston Johnson
Principal Sales Engineer
Condition Monitoring Systems
Vibration Analyst III - www.vibinst.org, www.mobiusinstitute.com
National Instruments
[email protected]
www.ni.com/mcm
www.ni.com/soundandvibration
www.ni.com/biganalogdata
512-683-5444

9233: Vibration signal contains dc offset

Hai all,
Am using cRIO 9012 controller and 9233 vibraiton module for acquiring vibration data from vehicle using IEPE accelerometers. Data iam acquiring is not symmetric about zero and the FFT of this acquired data shows a dominent peak (than any other peak) @ .1hz which means that signal has a dc offset. But from the specs 9233 has an ac coupling which means that dc signal should be filetered. Am getting problem in analysis due this behavious. Why is this behaviour? Can anyone help me out in this.
Any quick response will be appreciated.
With best regards,
JK
With regards,
JK
(Certified LabVIEW Developer)
Give Kudos for Good Answers, and Mark it a solution if your problem is solved.

Hi JK,
The DC offset comes from the amplifiers and the ADC.
You shouldn't have any useful data below 5Hz, so you can just ignore any bins that are 5Hz and below. For idle channel noise tests (and others) we ignore any bins that are 20Hz and below.
The DC offset is expected and you should ignore your really low frequency bins. Out of curiosity, why are you worried about the 0.1Hz bin anyway?
Hopefully this helps!!
Aashish M
Applications Engineer
National Instruments
http://www.ni.com/support/

Continuous wavelet transform of vibrational signals (scalogram)

Similar Messages

Maybe you are looking for