Development of Autonomous Smart Sensor Nodes for Hybrid Structural Health Monitoring of Large Structures

Abstract

본 연구에서는 대형구조물 하이브리드 구조건전성 모니터링을 위한 독립형 (autonomous) 스마트 센서 노드가 개발되었다. 이와 같은 연구 목표를 달성하기 위하여 다음과 같은 연구가 수행되었다. 첫째, 가속도 기반 스마트 센서 노드와 임피던스 기반 스마트 센서 노드를 개발하였다. 개발된 스마트 센서 노드는 (1) 마이크로 컨트롤러가 내장되어 있고, (2) 센싱 능력을 가지며, (3) 무선 통신이 가능하고, (4) 배터리로 동작하며, (5) 저가형으로 설계되었다. 다음으로 가속도 기반 스마트 센서 노드를 이용하는 전역적 구조건전성 모니터링, 임피던스 기반 스마트 센서 노드를 이용하는 국부적 독립형 구조건전성 모니터링과 가속도 및 임피던스 기반 스마트 센서 노드들을 상호보완적으로 이용하는 하이브리드 구조건전성 모니터링 전략을 제안하였다. 둘째, 스마트 센서 노드의 독자적인 구조건전성 모니터링을 위해 가속도 및 임피던스 특징 추출 과정이 설계되었다. 이를 위해, 특징 추출 과정에서 발생되는 오류들에 대해서 분석하고 그 해결 방안을 제시하였다. 특히, 센서 노드간의 시간 비동기화에 의한 오류를 극복하기 위해 복소수로 표현되는 모드형상을 이용하는 새로운 모드해석기법을 제안하였으며, 수치 및 모형 실험을 통해 그 유용성을 평가하였다. 셋째, 독립형 스마트 센서 노드의 성능을 평가하기 위한 대상구조물로서 PSC 거더 교량을 선정하고, 그에 적합한 구조건전성 모니터링 기법들을 선정하였다. 넷째, 각각의 선정된 구조건전성 모니터링 기법들을 가속도 및 임피던스 기반 스마트 센서 노드에 내장하여 일련의 손상 시나리오에 따라 스마트 센서 노드의 성능을 평가하였다. 마지막으로, 앞서 수행된 스마트 센서 노드의 성능 평가 결과를 바탕으로, 손상의 경보, 손상의 유형 판별, 손상 위치 및 크기를 식별할 수 있는 하이브리드 구조건전성 모니터링 전략을 설계하였으며, 그 성능을 실험을 통해 검증하였다.

In this study, autonomous smart sensor nodes for hybrid structural health monitoring (SHM) of large structures were developed. In order to achieve the objective, the following approaches were implemented. Firstly, acceleration-based and impedance-based smart sensor nodes were designed for global and local SHM. The smart sensor nodes have the following five essential features: 1) on-board microprocessor, 2) sensing capability, 3) wireless communication, 4) battery power, and 5) low-cost. The acceleration-based smart sensor node includes microcontroller, wireless radio, and signal processing units. The impedance-based smart sensor node includes microcontroller, wireless radio, and impedance chip. Then autonomous operation schemes for global and local SHM were designed on the basis of the decentralized wireless sensor networks. Also, two distinct autonomous operation logics for the hybrid SHM complementing global and local SHM methods were proposed as follows: 1) a serial operation logic and 2) a parallel operation logics. Secondly, the feature extraction process of the smart sensor nodes was designed for the autonomous SHM. Six vibration features including acceleration timehistory, frequency response functions, natural frequencies, mode shapes, modal strain energies, and modal flexibilities and two impedance features including impedance and admittance were briefly described. Then the feature extraction process for the autonomous SHM and various errors (such as noise, temperature effect, computational error, data-loss, and time-unsynchronization) in the feature extraction process were described. To reduce the error due to timeunsynchronization among the errors, a new modal analysis approach based on the complex mode-shapes was proposed. The performance of the approach was evaluated by performing numerical and experimental tests. From the evaluations, the mode shapes with small errors for time-unsynchronized signals were extracted by the proposed approach. Thirdly, the SHM methods were selected for autonomous SHM using the acceleration and impedance-based smart sensor nodes. Since the selection of SHM methods for the smart sensor nodes should be based on target structures, their inherent damage types, and the variety of extracted features, prestressed concrete (PSC) girder bridges were selected as the target structure. Among the promising SHM methods, eight vibration-based SHM methods and three impedance-based SHM methods were selected for the autonomous damage monitoring in the PSC bridges. Fourthly, autonomous SHM procedures using the smart sensor nodes were designed for global and local SHM of PSC girder bridges. To verify the proposed autonomous SHM procedures using the smart sensor nodes, a lab-scaled PSC girder model was selected. To implement the autonomous SHM, the global and the local SHM methods suitable for the PSC girder model were embedded in the smart sensor nodes. Then the performance of the autonomous SHM procedures was evaluated by performing experimental tests with the smart sensor nodes on the PSC girder model. Finally, an autonomous SHM operation scheme using the smart sensor nodes was designed for hybrid SHM of PSC girder bridges. For the hybrid SHM using the autonomous smart sensor nodes in PSC girder bridges, the appropriate SHM methods were selected from the evaluation tasks for the global and the local SHM using the smart sensor nodes. Then the parallel operation logic is adopted to design a hybrid SHM strategy for five mixed damage scenarios in the PSC girder model. By implementing the autonomous smart sensor nodes, the hybrid SHM strategy was evaluated by performing experimental tests on the PSC girder model. For two damage cases which have only girder damages, the damages were successfully identified by the smart sensor nodes. For two damage cases which have only tendon damages, the tendon damages were estimated with good accuracy while the girder damages were also indicated even though there is no girder damage (i.e., false positive error). For the damage case which mixed damages in girder and tendon, the girder damage was indicated with good accuracy but the tendon damage was estimated with very large error. In addition, for the damage cases which have mixed damages in internal and external tendons, the damages were estimated with small errors by the smart sensor nodes. For the practical application of smart sensor nodes, several subjects are recommended for further study. First, the smart sensor node needs to develop power harvesting strategies for long-time monitoring of structures. The smart sensor units can continuously operate during a few hours by battery power. Second, to monitor very large structures, wireless SHM system may need a multi-hop wireless sensor network. The multi-hop wireless sensor network may lead to improvement of scalability. Third, for the system to be used for years, the reliability of smart sensor nodes needs to be improved. Finally, in field applications of SHM, the smart sensor nodes may need multi-functional sensing strategies to measure other types of signal, temperature, humidity and wind speed. SHM strategy using multi-functional sensing (e.g., hybrid SHM in this study) shows the promising performance for SHM.