| The focus in this thesis is on control problems characterized by mathematical models which exhibit significant nonlinearity and uncertainty. The directions of this research are aimed at both the theory and application of nonlinear control systems.; First, initial steps are taken towards the development of a nonlinear adaptive control theory. A feedback linearization design is presented with the effects of both parametric and dynamic uncertainties included. An adaptive update law is proposed to counteract parametric uncertainties, and the robustness of the system to dynamic uncertainties is analyzed. In support of the new design methodology, a conceptually simple stability criterion is provided.; Second, contemporary feedback techniques are used to design nonlinear controls for a class of electromechanical systems. The emphasis lies on achieving improved dynamic performance from switched reluctance motors and direct drive robots, both characterized by physical simplicity at the cost of demanding control requirements. Nonlinear state feedback transformations are designed by the techniques of feedback linearization and composite control. These transformations ideally convert the system into a linear controllable one, providing decoupled controls for the stator voltages and robot links.; Third, a prototype implementation of a nonlinear feedback control system for three-phase switched reluctance motors is described. The prototype develops a smooth torque by driving the stator currents to a state dependent manifold upon which the closed loop system responds like a second-order, single-input linear system. The novel characteristics of the implementation are the high-speed linearizing and decoupling transformation circuit, which is programmable via a memory chip, and the automated nonlinear modeling software used to program this circuit. |
