Application of time series analysis for bridge monitoring

Abstract

Despite the recent considerable advances in structural health monitoring (SHM) of civil infrastructure, converting large amount of data from SHM systems into usable information and knowledge remains a great challenge. This paper addresses the problem through analysis of time histories of static strain data recorded by an SHM system installed in a major bridge structure and operating continuously for a long time. The reported study formulates a vector seasonal autoregressive integrated moving average (ARIMA) model for the recorded strain signals. The coefficients of the ARIMA model are allowed to vary with time and are identified using an adaptive Kalman filter. The proposed method has been used for analysis of the signals recorded during construction and service life of the bridge. By observing various changes in the ARIMA model coefficients, unusual events as well as structural change or damage sustained by the structure can be revealed.

Predicting thermal response of bridges using regression models derived from measurement histories

“…Existing data-driven approaches focus on the analysis of response measurements and ignore distributed temperature measurements. Omenzetter and Brownjohn [16] applied autoregressive integrated moving average models (ARIMA) to analyze strain histories from a full-scale bridge and noted that performance of the models could potentially be improved by including temperature measurements. More recently, Posenato et al [14] exploited the correlations between response measurements due to seasonal temperature variations for anomaly detection.…”

Lessons from monitoring the performance of highway bridges

“…For example, one of the topics on the agenda is the comparison of the efficiency of the two above methods. Also, the two methods, while reasonably successful, are not flawless as they may yield some spurious observations caused by non-Gaussian distribution of the strain data, which was discussed in detail elsewhere [30]. Overcoming successfully the methods' limitations is critical for making them potentially useful to the bridge operators.…”

“…The second analytical procedure operates directly on the strain time series and does not involve wavelet transform [30]. It was inspired by the studies of Sohn et al [31,32], who modeled dynamic signals using autoregressive (AR) time series models, and through examination of the changes in AR model parameters were able to detect damage.…”

A State of the Art Review of Modal-Based Damage Detection in Bridges: Development, Challenges, and Solutions

“…Given the present paper's aim and scope, the reader is advised to inquire the following publications for a more detailed account of AR type models' theoretical base [159,160]. However, it is important to note that their damage sensitivity arises from the assumption of stationarity, which allows for the detection of nonlinear behaviour in the vibration data by inducing changes in the residual errors and/or model parameters [161].…”