STFT methods

This is also a fairly straightforward method. Perform a Short Time Fourier Transform 8.1 (i.e. split the vibration signal up into segments and perform an FFT on each segment), and then for each segment pull out the amplitude of the bin which is at the frequency that corresponds to the shaft speed (or 2/rev, or whatever). Despite the simplicity, I generally do not recommend this method for many reasons:

• In a STFT, you have to deal with a time-frequency tradeoff. E.g. If you want to get the amplitude ten times per second, you will need frequency bins that are 10 Hz wide (assuming no overlap). That may be too wide of a passband for your application.
• This method will be susceptible to the ``picket fence effect'' 6.4.2 unless you use a flattop window. But if you use a flattop window then you need a lot of overlap to avoid throwing away data, which leads to increased processing time.
• The FFT frequency bins are a constant frequency bandwidth which means that at different speeds you get different order bandwidths. This can be a disadvantage if your speeds change signficantly.
• FFTs require more CPU time than other methods
• You don't get any output until you have collected an entire segment. E.g. if you have 1s long segments and there is a step change in the signal after 0.1s, you get no indication of this until another 0.9s has passed and you've collected the entire segment.

These various disadvantages can sometimes be mitigated if you are post-processing data after the fact and you can play around with different settings until you get a good answer. But the disadvantages are usually exacerbated for real-time monitoring.

The one advantage of the STFT method is if you want to track multiple different harmonics simultaneously. E.g. if you want to get the 1/rev and the 2/rev and the 3/rev and the 4/rev, etc., etc., you can get all of them in one computation. With some of the other methods you would have to repeat the entire calculation again for each harmonic to track.

The resample method described previously, requires no time-frequency tradeoff, has no picket fence effect, requires less processing power, and allows constant order bandwidth. Therefore I would recommend that as a better method to start with.

Figure 34:

" . $row['Name'] . " Posted on " . $row['DateTime']; echo "
"; echo $row['Comment']; echo "

"; } echo ""; ?> Leave a comment on this page:

Name: (optional)
To prove you are not a robot, what is 2+3?

This work by Daniel Kiracofe (daniel dot kiracofe at gmail dot com) is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License./' $I