Date of Award

Spring 1998

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mechanical & Aerospace Engineering

Program/Concentration

Engineering Mechanics

Committee Director

Jen-Kuang Huang

Committee Member

Chuh Mei

Committee Member

Sebastian Bawab

Committee Member

Keith M. Williamson

Abstract

This dissertation first presents indirect closed-loop system identification through residual whitening, then identifies the cavity noise system and applies controllers to reduce the noise. High speed air flow over the cavity produces a complex oscillatory flow-field and induces pressure oscillations within the cavity. The existence of cavities induces large pressure fluctuations which generate undesirable and loud noise. This may have an adverse effect on the objects, such as reducing the stability and performance of overall system, or damaging the sensitive instruments within the cavity.

System identification is the process of building mathematical models of dynamical systems based on the available input and output data from the systems. The indirect system identification by residual whitening is used to improve the accuracy of the identification result, and the optimal properties of the Kalman filter could be enforced for a finite set of data through residual whitening. Linear Quadratic Gaussian (LQG) and deadbeat controllers are applied to obtain the desired system performance. Linear Quadratic Gaussian (LQG) control design is the technique of combining the linear quadratic regulator (LQR) and Kalman tilter together, namely, state feedback (LQR) and state estimation (Kalman filter). Deadbeat control design is to bring the output to zero, and both indirect and direct algorithms are applied. For the indirect method, one needs to calculate the finite difference model coefficient parameters first, then form the control parameters. In the recursive direct algorithms, however, one can compute the control parameters directly. When systems change with time, the system parameters become time-varying. An adaptive predictive control is needed for this situation. Since the system parameters are time-varying, the control parameters need to be updated in order to catch up with the systems' changes. The classical recursive least-squares technique is used for the recursive deadbeat controller, and it could be operated for on-line application.

DOI

10.25777/ytat-gn68

ISBN

9780591815801

Share

COinS