Constrained maximum likelihood estimation for state space sampled-data models

Abstract:

The Expectation-Maximization algorithm is applied in this paper to estimate state-space sampled-data models including constraints on the location of the poles. Linear quadratic matrix inequalities are used as constraints to obtain a model that preserves properties of the continuous time system, such as stability or damping characteristics. The results of the algorithm are shown in a simulation study.

Año de publicación:

2018

Keywords:

• linear matrix inequalities
• Expectation-maximization
• Constraints
• system identification
• Maximum likelihood

Fuente:

scopus