System Identification of Dynamically Sensitive Structures using Wireless Smart Sensor Networks

点击翻译按钮获取中文摘要

Large stocks of buildings and bridges in North America and elsewhere are nearing their design life, or have already surpassed it. Many are vulnerable to dynamic effects (e.g., earthquakes), as their risk was initially under-estimated. We need to assess their current state in order to make sound engineering decisions on whether to replace, rehabilitate or retrofit them, and to predict their remaining life. This is possible through system identification, which combines sensor measurements obtained from the structure and mathematical procedures to yield information regarding its current state. An expensive exercise just a few years ago, wireless sensors have now made system identification on large-scale structures more affordable. There is no need for expensive wiring, and local processing capabilities on these sensors allow for hierarchical implementation of algorithms. Nevertheless, the advances in wireless sensor hardware have far out-paced our ability to effectively use and interrogate measured data. This is because, existing algorithms are sensitive to noise, not autonomous and demand large transmission loads. The main aim of this proposal is to overcome these drawbacks by developing a new class of algorithms. These algorithms will be decentralized, computationally light and autonomous, and will employ low transmission rates. This will be achieved using two powerful concepts in signal processing: time-frequency analysis and compressive sensing. The former will enable autonomous and decentralized implementations, while the latter will help us reduce data transmission. Theoretical, laboratory and field studies are proposed. This approach will lead to effective and affordable means of condition assessment to address the problem of aging infrastructure. Students will be trained in inter-disciplinary skills, which are in demand in Canada.