Collocation of sensing, computing, and actuation in low-power wireless nodes for smart structure applications in civil and mechanical systems

Effective management of civil structures requires both data and a physical means to mitigate the negative consequences of the effects of extreme loading events. This thesis presents a smart structure framework characterized by low-cost wireless nodes with collocated sensing, computing, and actuation capabilities. These nodes are intended to function as an automated first line of defense during extreme loading events, providing both rapid assessment of structural condition (i.e., health) and automated response (i.e., control). Low-cost wireless sensing and actuation nodes promote dense instrumentations that can provide great insight into the dynamic behavior and condition of structures. However, wireless networks should not be viewed merely as one-to-one replacements for traditional tethered systems. Rather, the goal of this thesis is to demonstrate the embedment of computationally expedient approaches for traditional smart structure tasks (i.e., load estimation, structural health monitoring, and structural control) implemented within wireless sensor and actuation networks. The distributed nature of these computing resources, coupled with limitations on power and communication bandwidth, require unique decentralized data processing algorithms that can operate effectively within the decentralized wireless smart structure environment. To accomplish this goal, this thesis first presents the development and validation of a novel wireless sensing and actuation platform necessary to meet the specific requirements of this thesis work. Then, using this wireless system, a method for estimating wind loading from measured wind turbine tower response is experimentally validated. This method can generate reference loading data that may be used to improve the design economy of future turbines. In addition, a wireless structural health monitoring method based on a physical parameterization of time-series model coefficients is presented for damage detection in post-earthquake scenarios. This method employs a physics-based method of evaluating and integrating damage indications derived from individual sensors within the network. Finally, a partially-decentralized method for wireless structural control is presented in which the wireless network dynamically trades bandwidth for performance of actuators engaged in feedback control. This method provides a means to allocate scarce bandwidth resources while still allowing the wireless controllers to improve performance by identifying and broadcasting only the most valuable feedback data over the network.