In my work, I use LMS Test-lab, in order to idetify modal parameters of a machine tool.
For exemple, I use the PolyMax method. In the stabilization diagram I always find difficulties to choose poles.
Taking for exemple, this stabilization diagram:
Here, I choose poles in this table:
Here, we found a pole on 500.97 HZ and on 545.22 Hz. I don't know if both of these two poles are structural or not or one is structural and the other is not. The same problem with 1252.58 Hz and 1271.37 Hz.
Also, the stabilization diagram change from a model size to another, which can influence on the identified modal base of the structure.
Is there a solution for the choice of the modal size?
Thank you a lot
An interesting topic. One observation: looking a your stabiliazation diagram, it does not look stable to me at all of the resonances. I would change the model size higher than 32 (at the bottom of the Stabilization sheet) and then press the Play Arrow in the bottom right corner to recalculate the Stab. Diagram. Do this until all the poles stabilize.
Also, the mode indicator function (MIF) is more helpful at identifying resonances sometimes, so I would turn that on as well. If you believe you have repeated roots, or resonances very close in frequency you would need at least two references so you can calculate two MIFs. For example, near 721 Hz is that one mode with frequency shifts due to mass loading (or temperature effects or bungee stretch) or is that two resonances? If it's a symetric structure, you will have repeated roots. It looks like if you increase the model size two poles may stabilize near 721 Hz. Without two references and two MIF's to look at though, you do not know if it is one resonance or two - the FRFs will not show you that, only the MIFs.
I usually increase the model size until my poles stabilize and then I tend to select the 2nd or 3rd "s" after the pole is stable. By by going up or down the line of stable "s" you may see a change in the FRF Synthsis so choose a good pole "s" that provides a good mode shape, a good FRF Synthesis and does not increase computation time too much by selecting an "s" at a model size much higher than you need.