John Watkins

/spl Hscr//sub /spl infin// active noise control of fan noise in an acoustic duct

In this paper, the problem of reducing fan related noise in an acoustic duct is considered. By installing magnetic bearings on the noise producing machinery, it is possible to collocate the anti-noise source with the disturbance noise source. This approach allows for global noise reduction through out the duct system. Using /spl Hscr//sub /spl infin// control theory, an active noise controller is designed that attains broadband as well as tonal noise reduction at all points along the duct. The controller design is based on a state space model identified from an infinite dimensional physical model. Simulation results demonstrate the global nature of this novel active noise control approach.

A Novel Approach to a Control Systems Laboratory

In controls education today, a significant gap exists between the material covered in the typical undergraduate classroom and the skills that students need to be practicing control system engineers. In order to help bridge this gap, a control systems laboratory was developed in the Systems Engineering Department at the United States Naval Academy (USNA) with the following design objectives. The first objective was to provide the students the opportunity to apply control theory to physical systems. The second objective was to develop labs where each student works through the complete control system design process. The third objective was to increase the students’ exposure to sampled-data control. The paper begins with a discussion of the Quanser rapid control prototype development system and laboratory experiments. Modeling and system identification are discussed next. Key areas that areas that are emphasized include the use of a dynamic signal analyzer for frequency domain identification and the identification of Coulomb friction for simulation purposes. A unified approach for root locus and Bode design that is used through out the course is discussed next. Finally, analog and digital controller implementations are discussed.Copyright

Control of time-delayed double integrator systems

The control of system with large time-delays is a very challenging problem. As Proportional-Integral-Derivative (PID) controllers and variants of these such as P, PI, or PD are widely used in industry, there has been extensive work to determine the range of their gains that will stabilize a linear time-invariant plant described by a rational transfer function. However, the extension of this work to systems with time-delays has been difficult. In this paper, the control of double integrators, a particular class of secondorder systems, with time-delays will be considered. Double integrator systems are a simple, but important, class of second order systems, as they model single-degree of freedom translational and rotational systems. The study of time-delayed double integrator systems in this paper is motivated by the formation flying challenge. In this paper, the ability for tight formation control in the presence of communication and measurement delays is investigated. The paper begins with a discussion of the cross-link requirements for formation control. A simple two degreeof-freedom model with communication delays is introduced for analysis. A typical approach for systems with small time-delays is to base the design on a nominal system model that does not contain the time-delays. The limitations of this approach on stability and closed-loop bandwidth are discussed. The time-delay is explicitly considered in the design of state-feedback (PD) controllers. The effects of the time-delay on stability, stability margins, and closed-loop bandwidth are investigated.

A MATLAB graphical user interface for linear quadratic control design

The optimal linear quadratic regulator (LQR) method is a powerful technique for designing controllers for complex systems that have stringent performance requirements. For most realistic applications, the LQR problem must be solved via a computer aided design (CAD) package such as MATLAB. With the CAD packages solving the optimization problems, the challenge lies in how the weighting matrices are chosen. With experience, practising engineers become quite adept at choosing these matrices for a specific application. Students in a typical quarter or semester long course do not have the luxury of this experience. Consequently, the authors have written a MATLAB graphical user interface (GUI) that allows the students to easily tune the weighting matrices and immediately see the consequences of those changes. This paper describes the motivation behind the LQR Design GUI, the GUI construction process, and the use of the LQR Design GUI in the senior-level modern control systems course at the U.S. Naval Academy.

On the use of magnetic bearings for global control of fan noise

In this paper, a novel approach is presented to reduce fan related noise in an acoustic duct. By installing magnetic bearings on the noise producing machinery, it is possible to collocate the anti-noise source with the disturbance noise source. This approach allows for global noise reduction throughout the duct system.

Adaptive auto-balancing control of magnetic bearings for an optical chopper

We show how adaptive notch filters and magnetic bearings can be utilized for auto-balancing of an optical chopper. This optical chopper is required. to operate between 4800 and 5200 RPM (rotations per minute) for five years in space. Life testing of the ball bearings in the optical chopper has indicated potential failure of the bearings after three to four years due to lubrication break down. The goal of this research is to investigate the feasibility of replacing the ball bearings with magnetic bearings. By using magnetic bearings to suspend the optical chopper, friction between the bearings and the rotor, and associated failures, can be eliminated. Initial testing using standard feedback controllers, however, showed excessive wobble in the chopper due to mechanical imbalances. By using adaptive notch filters based on the LMS (least mean squares) algorithm, we show how this wobble can be significantly attenuated. Experimental results are presented.