Synthesis of feedforward/feedback control systems for nonlinear processes.

Abstract

In this thesis, the unified problem of disturbance rejection and output tracking for general nonlinear processes is studied, using methods from differential geometry. An analysis framework is initially established, through a detailed study of the concept of relative order. The general problem of disturbance rejection and output tracking is formulated as a feedforward/feedback control problem, and is addressed first for single-input single-output processes and then for multiple-input multiple-output processes. Feedforward/state feedback laws are synthesized that completely eliminate the effect of measured disturbances on the controlled outputs and induce a well-characterized linear input/output behavior. A general feedforward/feedback control structure is developed that also accounts for modeling error and unmeasured disturbances. The developed control methodology is applied to composition control in a cascade of chemical reactors and to temperature and number average molecular weight control in a continuous polymerization reactor. On the basis of the properties of relative order and the controller synthesis results, the problem of synthesis of control configurations is also addressed. A general framework for the structural evaluation of alternative control configurations is developed, based on fundamental structural limitations in the control quality and structural coupling considerations. The developed evaluation framework is applied to the synthesis of control configurations in an evaporation unit, a continuous chemical reactor and a heat-exchanger network.