Design of self-repairing digital PID controllers for non-square multivariable plants

Othman, MZ

Abstract

The complexity of today's multivariable plants gives rise to the need for sophisticated control
systems to ensure high-performance operation whilst maintaining high system integrity. The
integrity of such systems is to be interpreted as the sensitivity of high-performance control
systems to failure and their capability for reconfiguration if necessary. Thus, the issue of
self-repairing control systems becomes practically very important. Indeed, besides simplicity
and practical realism, the associated controller of the self-repairing control system should
avoid the need for detailed mathematical models of the plant and should utilise only data
obtained from direct input-output measurements. Moreover, the self-repairing controller
should have the capability of altering its control law to promote plant survivability in the
face of severe plant-parameter variations characterised by actuator failure.
It is accordingly shown in this thesis that self-repairing digital control systems can be
designed by extending the domain of applicability of the digital PID controller introduced
by Porter et al (1985) so as to incorporate non-square plants subject to actuator failure.
This demonstration is effected by classifying the time-domain characteristics of non-square
linear multivariable plants using step-response matrices. These characteristics are used firstly
to design non-adaptive signal-following systems for non-square linear multivariable plants,
and then to design non-adaptive model-folio wing systems for such plants. In order to
produce self-repairing controllers, these non-adaptive controllers are then rendered adaptive
so that actuator failures can be tolerated.
The effectiveness of such self-repairing digital controllers is illustrated by designing digital
PID controllers for a three-input/two-output gas turbine, the three-input/two-output X-29
technology demonstrator aircraft, and a four-input/two-output two-link manipulator subject
to actuator failure.

• Thesis Type: Thesis
• Deposit Date: Jul 2, 2021
• Award Date: May 1, 1989