Thesis

Restricted structure non-linear generalized minimum variance control

Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2022
Thesis identifier
  • T16154
Person Identifier (Local)
  • 201652954
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • This research presents the Restricted Structure Non-linear Generalized Minimum Variance (RS-NGMV) algorithm for Linear Parameter-Varying (LPV) systems. The LPV systems are defined as linear plant subsystems within the control diagram and may include Non-linear (NL) input subsystems. The RS-NGMV control solution for the latter will be slightly different than the first one and have the capability of dealing with NL characteristics such as saturation, discontinuities and black-box terms. The controller is built in a low-order Restricted Structure (RS) in the form of a general z-transfer function. This brings forward two major advantages. First, it offers a high-order advanced control solution inside low-order control structures which are known for their natural robustness. Secondly, it is easier to operate and re-tune for the classically trained staff in the industry as it can be given the structures they are rather familiar with such as the PID. Another advantage of the RS-NGMV is its model-based design that enables a faster adaptation to implement different systems. Features of the RS-NGMV are investigated throughout the thesis with case studies from trends in engineering like robotics, autonomous and electric vehicles. The results show that the RS-NGMV is highly capable of adapting to set-point changes, parameter variations with its ability to update the control gains rapidly by using optimizations. Some extensions of algorithms have also been studied following recent directions in optimal/predictive control resulting in a new preview control approach and Scheduled RS-NGMV control.
Advisor / supervisor
  • Grimble, Mike
Resource Type
DOI

Las relaciones

Elementos