Teaching the student common methods to identify static and dynamical model in science, tecnology and industrial applications, their intrisic limits, and the trade-off between complexity and good fitting of the observed phenomenon.
Static and dynamic models for data: a-priori knowledge and measurement. Theory of prediction. Regression algorithms: the least-squares method and min-max optimization. Overfitting and countermeasures: regularization. Parametric estimation and internal model structures. Statistical analysis of parametric estimators. Bias in estimation and how to counteract it: maximum likelihood, instrumental variables. How to represent dynamics: AR, ARX, ARMA, ARMAX models and their identification; prediction form. Spectral analysis. Identification in closed loop.
For the detailed program, some teaching material and other information see the webpage:
Lennart Ljung. System Identification - Theory for the user, 2nd. ed. Prentice Hall.
Other teaching material will be supplied by the teacher.
Blackboard and chalk.
Written test and/or oral exam.
Other information will be made available on the teacher's public website,