Scalable Data Management and Data-Driven Analytics for Structural Condition Assessment using Structural Monitoring Data.

data management;data-driven analytics;structural condition assessment;structural monitoring data;Civil and Environmental Engineering;Engineering;Civil Engineering

Zhang, YilanLi, Victor C ;

University of Michigan

关键词: data management;    data-driven analytics;    structural condition assessment;    structural monitoring data;    Civil and Environmental Engineering;    Engineering;    Civil Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/135857/yilan_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Structural monitoring systems are an objective and quantitative-based management tool that have been developed to assist structure owners with their diagnostic and prognostic structural condition monitoring processes. As sensing technologies mature, the deployment of permanent sensing arrays in structures are becoming more popular resulting in increasing volumes of sensing data. This thesis focuses on the advancement of structural condition assessment by developing a scalable data management system for the curation and analysis of large-scale data sets associated with long-term structural monitoring systems.A hybrid database system termed SenStore is proposed consisting of a relational database system for the storage of structural information and a hierarchical data format (HDF) repository for the storage of time history sensor measurements. By storing sensor data in an HDF repository, sensor data can be queried at high speeds resulting in greater scalability of analytics to be performed on the data.The thesis also develops novel approaches to interrogating data associated with long-term structural monitoring systems.Data-driven analytics are built on top of the SenStore data management platform to extract information associated with the performance and condition of the monitored structure from the curated data.Automated modal parameter extraction by stochastic subspace identification (SSI) is implemented to obtain the modal properties of structures monitored using accelerometers.Long term trends in modal parameters are modeled by Gaussian process regression (GPR).A load estimation algorithm is also proposed for highway bridges monitored using strain gages.The data-driven analytics developed in the thesis are evaluated using long-term monitoring datasets from the New Carquinez Bridge (Vallejo, CA) and the Telegraph Road Bridge (Monroe, MI).