20810158 - Model Identification and Data Analysis

Introduce the student to the fundamentals of system identification applied to sampled systems (ARX and ARMAX model, ordinary least squares, recursive least squares, bayesian filtering). Introduce the student to sensor fusion. To familiarize the student with the use of the MatLab identification toolbox

Curriculum

scheda docente | materiale didattico

Programma

Dynamical models of stationary processes and prediction
- Physical laws in engineering and science
- Stochastic processes
- Models for filtering, prediction and control: Input-output models for time series and dynamical systems (AR, ARMA, ARX, ARMAX)

Identification
- Black-box identification (Least Squares and Maximum likelihood methods)
- Model complexity selection
- Cross-validation, FPE (Final Prediction Error), AIC (Akaike Information Criterion) or MDL (Minimum Description Length) techniques
- Recursive identification methods (RLS,ELS,RML). Adaptation via forgetting factor techniques

Bayesian filtering
- The state estimation problem. Filtering, prediction and smoothing.
- Kalman filter, steady-state filter Extended Kalman filter
- Unscented transformation, Unscented Kalman filter
- Grid-based filtering
- Particle filtering

Distributed filtering
- Information filter
- Extended Information filter

Testi Adottati

Sergio Bittanti, "Model Identification and Data Analysis", John Wiley and Sons Ltd

Bibliografia Di Riferimento

B.D.O. Anderson, J.B. Moore: Optimal filtering, Prentice Hall, 1979. Y. Bar-Shalom, X.R. Li, T. Kirubarajan: Estimation with applications to tracking and navigation, J. Wiley & Sons, 2001. B. Ristic, S. Arulampalam, N. Gordon: Beyond the Kalman filter: particle filters for tracking applications, Artech House, 2004.

Modalità Erogazione

Traditional

Modalità Frequenza

Not applicable

Modalità Valutazione

Written test, oral test.

scheda docente | materiale didattico

Programma

Dynamical models of stationary processes and prediction
- Physical laws in engineering and science
- Stochastic processes
- Models for filtering, prediction and control: Input-output models for time series and dynamical systems (AR, ARMA, ARX, ARMAX)

Identification
- Black-box identification (Least Squares and Maximum likelihood methods)
- Model complexity selection
- Cross-validation, FPE (Final Prediction Error), AIC (Akaike Information Criterion) or MDL (Minimum Description Length) techniques
- Recursive identification methods (RLS,ELS,RML). Adaptation via forgetting factor techniques

Bayesian filtering
- The state estimation problem. Filtering, prediction and smoothing.
- Kalman filter, steady-state filter Extended Kalman filter
- Unscented transformation, Unscented Kalman filter
- Grid-based filtering
- Particle filtering

Distributed filtering
- Information filter
- Extended Information filter

Testi Adottati

Sergio Bittanti, "Model Identification and Data Analysis", John Wiley and Sons Ltd

Bibliografia Di Riferimento

B.D.O. Anderson, J.B. Moore: Optimal filtering, Prentice Hall, 1979. Y. Bar-Shalom, X.R. Li, T. Kirubarajan: Estimation with applications to tracking and navigation, J. Wiley & Sons, 2001. B. Ristic, S. Arulampalam, N. Gordon: Beyond the Kalman filter: particle filters for tracking applications, Artech House, 2004.

Modalità Erogazione

Traditional

Modalità Frequenza

Not applicable

Modalità Valutazione

Written test, oral test.