The lecture is concerned with some fundamental concepts of parametric and non-parametric identification methods and with the design of optimal MIMO (multi-input-multi-output) state-space controllers and observers.

non-parametric identification methods (transient response analysis, fourier analysis ETFE), basics of stochastics, parametric identification methods (least squares with and without stochastic perturbation), model structures for identification (ARMA, ARX, ARMAX), Recursive Least Squares (RLS) method, Least Mean Squares (LMS) concept, weighted least squares method, optimal estimation (Gauß-Markov estimation, Minimum-Variance estimation), optimal observer (Kalman-Filter), dynamic programming due to Bellman, Linear Quadratic Regulator (LQR) problem with finite and infinite time horizon, optimal output regulator (Linear Quadratic Gaussian (LQG) problem), separation principle for the optimal controller and observer design, advanced concepts of state-space control (feedforward of the estimated perturbation, state-space controller with integral part, servoproblem)

oral exam

The registration via TISS is used for administrative purposes only.

Lecture Notes (in German) are available on the homepage of the institute. The link will be sent to the registered participants via News information.

This course is based on the concepts taught in VU Automatisierung. Furthermore, basic knowledge in linear algebra and stochastics is strongly recommended.