A Structural Health Monitoring Technique for the Analysis of Big Data of Bridges

Abstract

AbstractBig data (BD) in structural health monitoring for civil engineers has become possible because of recent advances in sensor networks, computing, information, and data acquisition systems technologies (SHM). The time-frequency analysis-based data-driven method provides good opportunities for implementing a strategy of long-term SHM for the bridge by using measured vibration signals. But there are limitations associated with complex and time-consuming feature extraction and decision-making procedures due to high-dimensional big data. Therefore, this paper proposes an innovative strategy that integrates the empirical mode decomposition (EMD) and Hilbert transform (HT) for informative feature extraction for bridge condition assessment. The efficiency of the suggested method is evaluated using data pertinent to a cable-stayed bridge. The findings demonstrate that the proposed technique is effective and robust in extracting informative features for bridge assessments in cases highly characterized by big data without any concerns regarding the information loss about the structural state. It is concluded that the proposed technique can address the issues of identifying data anomalies that are reliable early warning signs for future bridge failure and produce interpretable signal analyses that can handle BD with numerous periodic components, nonlinear functions, and periodic mode amplitudes as well as the problems of identifying data anomalies that are trustworthy early warning indications for upcoming bridge failure.KeywordsBig dataFeature extractionHealth monitoring data analytics