By power spectrum we are talking about an ASD (Auto Spectral Density) or PSD (Power Spectral Density). In fact PSD is often used when really ASD would be more correct.
Taking any signal and performing a frequency analysis using an ASD or a PSD will give the energy over a range of frequencies.
So what engineering decisions are made based on this data?
In vibration analysis, this data is used for resonance (or natural frequency) analysis. If a structure or component has a resonance at a particular frequency that is being excited by normal conditions then the part needs to be redesigned. This means either changing the mass or the stiffness to move the natural frequencies to another, higher frequency. The Auto Power Spectrum will show this.
When studying acoustics or vibration often transfer functions are used. These are usually, but not always, based on the auto power spectrum.
Studying transfer functions is very useful whether you are interested in noise, vibration or both. How much will this new damper reduce the vibration? How much will this new window reduce the noise from the building site? All these and more are analyses based on the measured power in the spectrum.
Latest posts by James Wren (see all)
- How Do I Upsample and Downsample My Data? - January 27, 2017
- What Are Vibration, Torsional Vibration & Shaft Twist? - November 8, 2016
- The Effect of the Nyquist Theory on Rotational Order Analysis - October 7, 2016
This Post Has 2 Comments
I was performing analysis on the vibration data using the auto spectral density function and I could not understand how to set the parameters like frequency spacing, percentage overlap, etc.
The Y-axis of the resultant graph had a drastic change in the value of the vibration when I was changing the frequency spacing from 0 to 1.
Can you help me to know what setting would give proper values in the auto spectral density function.
Thank you for posting on our blog.
Generally speaking when you set the frequency spacing to zero, some non-zero value is selected for you automatically by the software, as it is of course not possible to have zero spacing.
When I’m working in the frequency domain I tend to use 1Hz myself, but it depends on the frequency range and what your studying, there is no simple right or wrong value.
There will be changes in the amplitude when you select different spacing, for example if the spacing was very large, say 100Hz then if there was a peak at 50Hz, you would miss it.
If you then used a spacing of 50Hz you would see the peak appear.
So it’s important to understand what your software package is doing and to think about what you are doing and what your looking for.
The number one rule is that you should have some idea what to expect before you get your results.
Please feel free to ask if you have further questions at all.