| In this dissertation, the basic theory and properties of the Singular Value Decomposition (SVD) are presented for integral equation models of Distributed Parameter Systems (DPS) in the context of process identification and control. In addition, new methods of analysis and computation are described in which SVD-based techniques are used to provide a practical solution for the DPS identification and control problem.; Once developed, these novel procedures are then applied to a class of linear DPS which are conventionally described by parabolic partial differential equation models. An alternative integral equation representation for these processes is used to provide a general and readily identifiable pseudo-modal input/output model for a wide variety of such DPS, including both diffusive and convective/diffusive systems.; Results from numerical investigations concerning the parametric variation of certain properties related to this representation are presented and analyzed for several common boundary conditions. Detailed identification and control results are obtained for models of various physically interesting systems. Examples for the computer implementation of the algorithms developed in this work are given. Through these results, it is demonstrated that this SVD-based integral equation model of linear DPS gives a sound and convenient basis for the design of psuedo-modal feedback control systems.; Although these strategies are theoretically developed for a class of linear systems, it is shown that these methods may be extended to handle certain nonlinear DPS, such as a tubular chemical reactor model, by means of appropriate linearization techniques and through the use of model predictive control strategies such as Dynamic Matrix Control (DMC). Based on the results of these investigations, suggestions for future work are given. |
