Original scientific paper

Using orthonormal functions in model predictive control

Amila Dubravić ; Faculty of Electrical Engineering, University of Tuzla, Franjevačka 2, 75000 Tuzla, Bosnia and Herzegovina
Zenan Šehić ; Faculty of Electrical Engineering, University of Tuzla, Franjevačka 2, 75000 Tuzla, Bosnia and Herzegovina

Abstract

This paper describes use of Legendre orthonormal functions for representation of the control trajectory in discrete model predictive control, precisely, above mentioned functions are used for efficient parametrization of the difference of control signal. When high demands, fast sampling or complicated dynamics are present in the design, classical approach is not computationally efficient and can lead to poorly numerically conditioned solutions as noted in [1]. Using orthonormal functions, moreover Legendre functions, described in this paper, number of parameters used for description of future control trajectory is reduced and the trajectory itself becomes smoother, with control signal being of smaller amplitude then in the Laguerre case.

Keywords

model predictive control; orthonormal functions

Hrčak ID:

86723

URI

https://hrcak.srce.hr/86723

Publication date:

19.9.2012.

Article data in other languages: croatian

Visits: 2.643 *