Date of Award

2017

Document Type

Open Access Dissertation

Degree Name

Doctor of Philosophy in Civil Engineering (PhD)

Administrative Home Department

Department of Civil and Environmental Engineering

Advisor 1

R. Andrew Swartz

Committee Member 1

William Bulleit

Committee Member 2

Qingli Dai

Committee Member 3

Bo Chen

Abstract

Structural control is used to mitigate unwanted vibrations in structures when large excitations occur, such as high winds and earthquakes. To increase reliability and controllability in structural control applications, engineers are making use of semi-active control devices. Semi-active control gives engineers greater control authority over structural response versus passive controllers, but are less expensive and more reliable than active devices. However, the large numbers of actuators required for semi-active structural control networks introduce more cabling within control systems leading to increased cost.

Researchers are exploring the use of wireless technology for structural control to cut down on the installation cost associated with cabling. However wireless communication latency (time delays in data transmissions) can be a barrier to full acceptance of wireless technology for structural control. As the number of sensors in a control network grows, it becomes increasingly difficult to transmit all sensor data during a single control step over the fixed wireless bandwidth. Because control force calculations rely on accurate state measurements or estimates, the use of strategic bandwidth allocation becomes more necessary to provide good control performance. The traditional method for speeding up the control step in larger wireless networks is to spatially decentralize the network into multiple subnetworks, sacrificing communication for speed.

This dissertation seeks to provide an additional approach to address the issue of communication latency that may be an alternative, or even a supplement, to spatial decentralization of the control network. The proposed approach is to use temporal decentralization, or the decentralization of the control network over time, as opposed to space/location. Temporal decentralization is first presented with a means of selecting and evaluating different communication group sizes and wireless unit combinations for staggered temporal group communication that still provide highly accurate state estimates. It is found that, in staggered communication schemes, state estimation and control performance are affected by the network topology used at each time step with some sensor combinations providing more useful information than others. Sensor placement theory is used to form sensor groups that provide consistently high-quality output information to the network during each time step, but still utilize all sensors. If the demand for sensors to communicate data outweighs the available bandwidth, traditional temporal and spatial approaches are no longer feasible.

This dissertation examines and validates a dynamic approach for bandwidth allocation relying on an extended, autonomous and controller-aware, carrier sense multiple access with collision detection (CSMA/CD) protocol. Stochastic parameters are derived to strategically alter back-off times in the CSMA/CD algorithm based on nodal observability and output estimation error. Inspired by data fusion approaches, this second study presents two different methods for neighborhood state estimation using a dynamic form of measurement-only fusion. To validate these wireless structural control approaches, a small-scale experimental semi-active structural control testbed is developed that captures the important attributes of a full-scale structure.

Download

Share

COinS