Trajectory-scheduling control systems and their multi-objective design automation

Chong, Gregory Chow Ye (2006) Trajectory-scheduling control systems and their multi-objective design automation. PhD thesis, University of Glasgow.

Abstract

This thesis encompasses the analysis of TSN systems and their multi-objective design methods. TSN nodes are networked through interpolation and activation, similar to a gain-scheduling or local model/controller network. However, to achieve accuracy and ease of commissioning without requiring a large number of nodes, an algorithm has been developed first to identify optimum transition nodes within the entire operating envelope. Then the TSN approaches a nonlinear plant globally, not just locally, without requiring linearization. If desired or necessary, global optimisation provides an enhancement in the design process for TSNs. Since optimising only one aspect (a single objective) of performance while compromising others is undesirable, multi-objective designs have been developed concurrently to deliver or improve multiple aspects of performance.

Following the development of a TSN, it is applied to nonlinear system modelling, and this TSN is termed a Trajectory-Scheduling Model (TSM). A TSM possesses the same properties and design features as the TSN generic framework. A nonlinear system, a coupled liquid-tank, is used to examine this modelling technique. Results verify the feasibility and effectiveness of the methods developed and validates the TSM.

Further, the TSN technique is applied to nonlinear controller design, by way of a Trajectory-Scheduling Controller (TSC) network. It is illustrated through the design of a networked, easy-to-understand and easy-to-use PID control system for the coupled liquid-tank. Results show that the methods developed offer a high-performance linear control system with nonlinear capabilities to handle practical systems operating in a broad range and to cope with conflict between setpoint following at transient and disturbance rejection at steady state. This method is then applied to the PID network design problems for two nonlinear chemical processes.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: T Technology > TJ Mechanical engineering and machinery
• Colleges/Schools: College of Science and Engineering > School of Engineering
• Supervisor's Name: Supervisor, not known
• Date of Award: 2006
• Depositing User: Angi Shields
• Unique ID: glathesis:2006-3728
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 06 Nov 2012
• Last Modified: 10 Dec 2012 14:10
• URI: https://theses.gla.ac.uk/id/eprint/3728