Performing Bayesian system identification is hard:
Complex statistical methods
High computational costs
Programming skills
Then… why doing it?
✅ Theoretically consistent method to integrate measurement data with structural models
✅ Estimates the uncertainties of predictions rather than just giving a ‘best guess’
✅ Incorporates knowledge of the structure via the priors
Typical workflow¶
This how a typical workflow looks like:
