The course presents the main estimation and identification techniques to be used in the context of complex dynamic systems analysis, forecasting and control.
The learning outcomes of the course refer to the capacity of:
The course prerequisites refer to basic elements of systems theory, statistics and optimization.
Aula lessons and laboratory exercises.
Estimation theory: parameter estimation (correctness, consistency and efficiency of the estimator), Cramer-Rao theorem, minimum variance estimation (UMVUE and BLUE estimators), maximum likelyhood estimation, linear estimation with measurement errors (least square estimation, Gauss Markov estimator). Bayesian estimation (minimum squared error estimation and linear minimum squared error estimation). Kalman filter.
Identification techniques: definition of the parameter identification problem, model families (ARX, ARMAX, OE, ARXAR, BJ), MPE identification: convergence theorems, identification for ARX models (least squares identification), for ARMAX models and for ARXAR models, batch and iterative algorithms.
L. Ljung, "System Identification: Theory for the user", Prentice Hall (2nd Edition), 1999.
S.M. Kay, "Fundamentals of Statistical Signal Processing: Estimation Theory", Prentice Hall, 1993.
SIMONA SACONE (President)
MICHELA ROBBA
SILVIA SIRI (President Substitute)
The evaluation consists in an oral exam.
During the exam the student has to present the main arguments of the course, to solve numerical exercises and to explain the theoretical notions necessary for their solution
