qLPV predictive control : a benchmark study on state space vs input-output approach

This paper presents a comparison and evaluation of two approaches to Nonlinear Model Predictive Control (NMPC) via quasi-LPV modeling, by means of a benchmark problem: control of a 4 degree-of-freedom Control Moment Gyroscope (CMG). The use of quasi-LPV modeling allows us to recast the nonlinear optimization problem arising in NMPC, as a repeated quadratic program which can be solved efficiently. The difference between the two presented schemes lies in the modeling paradigm chosen to express the dynamics of the system, namely state space (SS) or input-output (IO) frameworks. In both cases, quasi-LPV models are obtained by performing a velocity-based linearization, which results in an exact representation of the nonlinear dynamics and enables offset free control. Both schemes are successfully implemented on a laboratory CMG, and the experimental results are compared and discussed. Furthermore, advantages and disadvantages of each control scheme are examined.

DDC Class

620: Ingenieurwissenschaften

Publication version

publishedVersion

Lizenz

https://creativecommons.org/licenses/by-nc-nd/4.0/

Name

1-s2.0-S2405896319322621-main.pdf

Size

1.04 MB

Format

Adobe PDF

Options

qLPV predictive control : a benchmark study on state space vs input-output approach