Nonlinear control and its application to active tilting-pad bearings

The drawbacks of active magnetic bearings are arousing interest in the adaptation of mechanical bearings for active use. A promising mechanical bearing candidate for active operation is the tilting-pad bearing. In this research, we introduce an active tilting-pad bearing with linear actuators that translate each pad. The use of feedback in determining the actuator forces allows for the automatic, continuous adjustment of the pad position during the machine operation. In this work, we develop the dynamic model of the active bearing system such that the actuator forces are the control inputs. The hydrodynamic force is modeled as a spring/damper-like force with unknown damping and stiffness coefficients. Whereas in the literature, the damping and stiffness effects are normally considered linear, here, motivated by a numerical study based on the Reynolds equation, we use a nonlinear model for the stiffness force. An adaptive controller is designed to asymptotically regulate the rotor to the bearing center. The proposed control design is applicable to both the linear and nonlinear stiffness models. Simulations and experiments show that the active strategy improves the bearing performance in comparison to its traditional passive operation. Further, the experiments indicate the nonlinear stiffness-based controller slightly improves the active bearing regulation performance relative to the linear-based one. To the best of our knowledge, this dissertation is the first to report the experimental demonstration of an active tilting-pad bearing using feedback control. Since the model of the active tilting-pad bearing has a parametric strict-feedback-like form, the second part of this dissertation is dedicated to constructing new nonlinear control tools for this class of systems. Specifically, we consider the regulation and tracking control problems for multi-input/multi-output parametric strict-feedback systems in the presence of additive, exogenous disturbances and parametric uncertainties. For such systems, robust adaptive controllers usually cannot ensure asymptotic tracking or even regulation. In this work, under the assumption the disturbances are C2 with bounded time derivatives; we present a new C0 robust adaptive control construction that guarantees the output/tracking error is asymptotically driven to zero. Numerical examples illustrate the main results, including cases where the disturbances do not satisfy the aforementioned assumptions