THE USE OF DYNAMIC MIXTURES IN ADVISING AND CONTROL

BÖHM, J.

Abstract

Mixtures were designed to model multimodal distributions. Typical application is the description of data located in cluster in the space. Similarly it is possible to use dynamic mixtures to model processes exhibiting several modes of behaviour. Normal dynamic mixtures are the extension of popular linear regression model. They can be used to describe systems working in several modes, typically nonlinear plants. Several optimization problems can be formulated for such plants. The paper presents the solutions to these problems and their applications.

Coresponding author e-mail: bohm[at]utia[dot]cas[dot]cz

Session: Linear and Non-linear Control System Design