| Popis: |
The methodology for the design of error-actuated digital set-point tracking controllers proposed by Porter and co-workers has emerged as a result of the pursuit of effective and practical solutions to the problem of designing digital control systems for unknown, dynamically complex multivariable plants with measurable outputs. In this thesis, such digital set-point tracking controllers and the resulting digital set-point tracking systems are enriched to embrace plants with unmeasurable outputs and plants with more outputs than manipulated inputs. In the study of the latter plants, the novel concepts of limit tracking (i.e. the tracking exhibited by plants with more outputs than inputs) is introduced and an associated methodology for the design of self-selecting controllers is proposed. Such controllers involve the selection of different set-point tracking controllers to control the most critical subset of plant outputs based upon the developed rigorous theoretical foundations for the limit-tracking systems. In such foundations, the classification of linear multivariable plants into Class I and Class II plants based upon their steady-state transfer function matrices facilitates the assessment of the feasibility of limit-tracking systems. Furthermore, the associated order-reduction technique simplifies the problem of deciding the minimum numbers of different subsets of plant outputs to be controlled by corresponding set-point tracking controllers. In addition, the dynamical properties of limit-tracking systems are also investigated using the phase-plane method and a methodology for the design of supervisory self-selecting controllers is proposed so as to prevent the occurrence of dynamical peculiarities such as limit-cycle oscillations which might happen in limit-tracking systems. The effectiveness of all the proposed methodologies and techniques is illustrated by examples, and the robustness properties of set-point tracking systems and limit-tracking systems in the face of plant variations and unknown disturbances are tested. Finally, self-selecting controllers are designed for a nonlinear gas-turbine engine and their practical effectiveness is clearly demonstrated. |
