Test and theory for robust active vibration control of rotor- bearing systems utilizing piezoelectric actuators

Abstract

This research focuses on designing effective control strategies using piezoelectric actuators for control of vibrations of flexible rotor-bearing systems. The actuators are completely described and the elements of feedback control are modeled and generalized closed-loop analysis methods are developed. These include incorporating controllers described in polynomial form in the s-domain. This allows for transfer functions of feedback elements obtained by analytical or experimental methods to be included in the closed-loop model in a direct fashion. Digital control is explored and effects unique to such systems are explained. The relation between continuous-time and discrete-time descriptions are also developed. Sampling effects are modeled so the closed-loop analysis may predict their effects on overall system performance. Decentralized control using PD-loops in both analog and digital domains is described in the context of a full-scale helicopter engine shaft-line. This long very flexible system poses a challenge to accurate modeling, and the combination of analytical and experimental measurements to establish model parameters are described. Results of control using a single plane of control establish the piezoelectric actuator as an effective device for vibration control. A new scheme of control involving a hybrid strategy of analog control in real-time coupled with a self-tuning scheme on a background computer is presented. By constructing a figure of merit based on a variant of the [infinity]-norm to establish the level of vibration in a flexible rotor, the feedback gains are optimized at a specific steady-state condition. During run-up and shutdown, the optimal gains are scheduled by interpolating in a linear fashion to provide effective control in the frequency range of operation. Such a scheme is most useful to account for changes in the system to be controlled or non-linearities which were not known or anticipated. A novel design methodology based on QFT is developed for AVC applications. This technique allows for establishing performance objectives as well as structured uncertainties in the system parameters. Both single-input single-output (SISO) and multi-input multi-output (MIMO) cases are considered. The process is illustrated for small-order systems. This is followed by an application of this theory to design a controller to control a flexible rotor-bearing system...