Structural Control Using Regenerative Force Actuation Networks

Citation

Scruggs, Jeffrey Thomas (2004) Structural Control Using Regenerative Force Actuation Networks. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/W3M9-ZW72. https://resolver.caltech.edu/CaltechETD:etd-06012004-063432

Abstract

A Regenerative Force Actuation (RFA) Network consists of multiple electromechanical forcing devices distributed throughout a structural system and actuated in such a way as to reduce the response of the structure when subject to an excitation. The associated electronics of the devices are connected together such that they are capable of sharing electrical power with each other. This makes it possible for some devices to extract mechanical energy from the structure, while others re-inject a portion of that energy back into the structure at other locations. The forcing capability of an RFA network is constrained only by the requirement that in the aggregate the total network must always dissipate energy.

The electromechanical currents generated by RFA networks must be controlled to create the desired structural forces. This control is facilitated by the alternation of a multitude of power-electronic transistor switches in the electrical network. In this study, a sliding-mode switching controller is proposed for realizing zero-error force command tracking. It is shown that parameter uncertainty is a critical issue for force commands which require the network to operate near its optimum transmissive efficiency.

RFA networks can be used to create velocity-proportional damping forces in structures. However, unlike traditional structural damping, RFA networks have the ability to create non-local and asymmetric damping forces. It is shown that this more generalized damping capability can lead to significant improvements in the forced response of a structure, as compared with traditional linear damping.

RFA networks may also be used for feedback control. In this context, the forcing capability of the RFA network is constrained by its physical limitations. In this study, a systematic method of nonlinear control design called "Damping-Reference" control is proposed, which guarantees a certain level of quadratic performance for the structural response. Variants of the control law synthesis are proposed for quadratic regulation, stochastic control, and H[infinity] control contexts.

These ideas are illustrated in the context of earthquake engineering through a simulation example, involving a three-story structure with a two-actuator RFA network installed. In this example, it is shown that the "power sharing" nature of the RFA network has a significant influence on the response.

Thesis Files