Automatic Measurement Selection

You need to focus on many things to get high-quality FRF models of your system. One of the tasks is selecting the individual FRF measurements to maximise the overall coherence in the data. This takes a lot of time and experienced engineering judgment – which can seem like witchcraft.

Here at VIBES, we feel you on this and developed Automatic Measurement Selection, or in short AMS. It allows you to objectively select the most coherent FRF in a second – no magic.

How to: Work with AMS

You can find AMS in Measure. When you run it, DIRAC applies AMS to the current excitation.

The basis for this selection is a new metric that we call measurement similarity. It is calculated for any subset of measurements of one excitation. Similarity considers the coherence of all sensor channels, so it’s best compared to a column of the green coherence matrix in Analyse. Simply speaking, what AMS does is finding the selection of measurements that makes the column “as green as possible”. Of course, you can also just apply AMS to all excitations, to optimise the whole coherence matrix at once.

Which measurements are selected depends on the settings of the AMS preset.

How to: Choose the AMS settings

You can specify different settings for each AMS preset. The similarity threshold lets you define a minimum similarity for the selected measurements. However, independent of this threshold, AMS will always select the two best-matching measurements. So, setting the threshold to 100% will still result in two measurements selected per excitation. When there are more measurements that meet the minimum similarity requirement, AMS will include as many as possible. But if you want to keep your file size small, you can set a maximum number of valid measurements, run AMS, and then delete all rejected measurements.

Another important setting is the frequency range in which AMS optimises similarity. By default, this is the entire frequency range which is specified in the DAQ settings. So, here, you can narrow it down to only your frequency range of interest.

If you use multiple hammers, AMS will only operate within the limits of the high, low, or band-pass filters that you defined for each hammer in the DAQ settings. By default, AMS is applied to all hammer groups, but the similarity is evaluated for each group separately. So, you always get at least the two best-matching measurements for each hammer. If you want to apply different AMS settings for each hammer, you can create one preset for each of them.

So far, we have only used AMS to select the most similar measurements of already recorded measurements. But you can also use it while measuring.

How to: Use AMS in live mode

When you double click on AMS, it is set in Live mode. Now, your personal measurement assistant DIRAC, will apply AMS as you are measuring and notify you with the sound feedback whether the impact meets the similarity requirement, or not.

And if you have set a maximum number of measurements, DIRAC will also notify you when you have enough good measurements, so you can move on to the next excitation point knowing that you have great measurement similarity.