Introduction

Welcome to the crash course on Bayesian system identification. This approach allows us to systematically combine:

• Physical/mathematical models of engineering systems

• Noisy sensor measurements

• Prior engineering knowledge and constraints

The result? Calibrated parameters with quantified uncertainty.

Course Structure¶

Through hands-on exercises, you’ll learn to:

1. Apply Bayes’ theorem for events

2. Use conjugate priors for analytical solutions

3. Perform Bayesian inference via numerical integration

4. Use the specialized Bayesian system identification library probeye

5. Handle multiple sensors and parameter

6. Apply advanced sampling methods (MCMC, Nested sampling)

7. Compare and select between competing models

Requirements¶

To follow along, you will need:

• A Python version between 3.6 and 3.11

• A code editor that supports Jupyter notebooks (e.g., VS Code)

• A virtual environment with the required packages installed: probeye, ipykernel