How to handle multiple discontinuous data sets.

[taylorh140]

Many experiments may only be performed in short operational bursts. To gain a better idea of the system behavior multiple runs can be used to help gain assurance, and provide better model fitting. Does this library support data, if it does how should the data be formatted?

[CPCLAB-UNIPI]

Dear Taylor,
we are not sure to understand well your question.
SIPPY is designed to take as input, data obtained from different sources (experimental set-up, machines, equipment, plants, or simulation).
Take always the user guide (here in attach) as a reference for the right data format.
If you want to run multiple identification analyses for repeated data sets of the same system, you can embed the call at SIPPY module in a for loop and store the different solutions for post-processing analysis.
Please let us know more in details what is your specific request and we could be more helpful.
Thanks,
The SYPPY team

[jamestjsp]

I think the question is "How to select data segments form different experiments?". Some methods have discussed in this book.

System Identification Theory for the User: Second Edition, Lennart Ljung, Linkoping, University Sweden

14.3 SELECTING SEGMENTS OF DATA AND MERGING EXPERIMENTS

When data from an identification experiment or in particular, from normal operating records are plotted, it often happens that there are portions of bad data or nonrelevant information. The reason could be that there are long stretches of missing data which will be difficult or computationally costly to reconstruct. There could be portions with disturbances that are considered to be non-representative, or that take the process into operating points that are of less interest. In particular for normal operating records, there could also be long periods of "no information:" nothing seems to happen that carries any information about the/process dynamics. In these cases it is natural to select segments of the original data set which are considered to contain relevant information about dynamics of interest. The procedure of how to select such segments will basically be subjective and will have to rely mostly upon intuition and process insights.