Limits...
Operational Modal Analysis of Bridge Structures with Data from GNSS/Accelerometer Measurements

View Article: PubMed Central - PubMed

ABSTRACT

Real-time dynamic displacement and acceleration responses of the main span section of the Tianjin Fumin Bridge in China under ambient excitation were tested using a Global Navigation Satellite System (GNSS) dynamic deformation monitoring system and an acceleration sensor vibration test system. Considering the close relationship between the GNSS multipath errors and measurement environment in combination with the noise reduction characteristics of different filtering algorithms, the researchers proposed an AFEC mixed filtering algorithm, which is an combination of autocorrelation function-based empirical mode decomposition (EMD) and Chebyshev mixed filtering to extract the real vibration displacement of the bridge structure after system error correction and filtering de-noising of signals collected by the GNSS. The proposed AFEC mixed filtering algorithm had high accuracy (1 mm) of real displacement at the elevation direction. Next, the traditional random decrement technique (used mainly for stationary random processes) was expanded to non-stationary random processes. Combining the expanded random decrement technique (RDT) and autoregressive moving average model (ARMA), the modal frequency of the bridge structural system was extracted using an expanded ARMA_RDT modal identification method, which was compared with the power spectrum analysis results of the acceleration signal and finite element analysis results. Identification results demonstrated that the proposed algorithm is applicable to analyze the dynamic displacement monitoring data of real bridge structures under ambient excitation and could identify the first five orders of the inherent frequencies of the structural system accurately. The identification error of the inherent frequency was smaller than 6%, indicating the high identification accuracy of the proposed algorithm. Furthermore, the GNSS dynamic deformation monitoring method can be used to monitor dynamic displacement and identify the modal parameters of bridge structures. The GNSS can monitor the working state of bridges effectively and accurately. Research results can provide references to evaluate the bearing capacity, safety performance, and durability of bridge structures during operation.

No MeSH data available.


GNSS-RTK signal before and after Chebyshev filtering.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5375722&req=5

sensors-17-00436-f009: GNSS-RTK signal before and after Chebyshev filtering.

Mentions: The real vibration data of the structure in the GNSS data were covered by noises due to the interferences caused by multipath errors and random noises. According to Section 3, the AFEC mixed filtering algorithm was applied to process the polluted GNSS data. First, an 8-order I-type Chebyshev high-pass filter was used to eliminate multipath errors (Figure 9). Second, the random noise error was reduced by the EMD based on autocorrelation function. The autocorrelation function of the signal reflects correlations at different times because of the statistical characteristics of the random signal. The autocorrelation function of random noise is characterized by a maximum at zero, a rapid decay to very little at other points. There is no such rule for the ordinary signal because the ordinary signal has a correlation between different moments. It is possible to distinguish whether the IMF component is a noise component by determining whether the autocorrelation function characteristic of the IMF component complies with the autocorrelation function characteristic of the random noise signal. Therefore, the random noise and ordinary signal can be distinguished by the autocorrelation function of signals. The autocorrelation function between the random noises and ordinary signal is presented in Figure 10.


Operational Modal Analysis of Bridge Structures with Data from GNSS/Accelerometer Measurements
GNSS-RTK signal before and after Chebyshev filtering.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5375722&req=5

sensors-17-00436-f009: GNSS-RTK signal before and after Chebyshev filtering.
Mentions: The real vibration data of the structure in the GNSS data were covered by noises due to the interferences caused by multipath errors and random noises. According to Section 3, the AFEC mixed filtering algorithm was applied to process the polluted GNSS data. First, an 8-order I-type Chebyshev high-pass filter was used to eliminate multipath errors (Figure 9). Second, the random noise error was reduced by the EMD based on autocorrelation function. The autocorrelation function of the signal reflects correlations at different times because of the statistical characteristics of the random signal. The autocorrelation function of random noise is characterized by a maximum at zero, a rapid decay to very little at other points. There is no such rule for the ordinary signal because the ordinary signal has a correlation between different moments. It is possible to distinguish whether the IMF component is a noise component by determining whether the autocorrelation function characteristic of the IMF component complies with the autocorrelation function characteristic of the random noise signal. Therefore, the random noise and ordinary signal can be distinguished by the autocorrelation function of signals. The autocorrelation function between the random noises and ordinary signal is presented in Figure 10.

View Article: PubMed Central - PubMed

ABSTRACT

Real-time dynamic displacement and acceleration responses of the main span section of the Tianjin Fumin Bridge in China under ambient excitation were tested using a Global Navigation Satellite System (GNSS) dynamic deformation monitoring system and an acceleration sensor vibration test system. Considering the close relationship between the GNSS multipath errors and measurement environment in combination with the noise reduction characteristics of different filtering algorithms, the researchers proposed an AFEC mixed filtering algorithm, which is an combination of autocorrelation function-based empirical mode decomposition (EMD) and Chebyshev mixed filtering to extract the real vibration displacement of the bridge structure after system error correction and filtering de-noising of signals collected by the GNSS. The proposed AFEC mixed filtering algorithm had high accuracy (1 mm) of real displacement at the elevation direction. Next, the traditional random decrement technique (used mainly for stationary random processes) was expanded to non-stationary random processes. Combining the expanded random decrement technique (RDT) and autoregressive moving average model (ARMA), the modal frequency of the bridge structural system was extracted using an expanded ARMA_RDT modal identification method, which was compared with the power spectrum analysis results of the acceleration signal and finite element analysis results. Identification results demonstrated that the proposed algorithm is applicable to analyze the dynamic displacement monitoring data of real bridge structures under ambient excitation and could identify the first five orders of the inherent frequencies of the structural system accurately. The identification error of the inherent frequency was smaller than 6%, indicating the high identification accuracy of the proposed algorithm. Furthermore, the GNSS dynamic deformation monitoring method can be used to monitor dynamic displacement and identify the modal parameters of bridge structures. The GNSS can monitor the working state of bridges effectively and accurately. Research results can provide references to evaluate the bearing capacity, safety performance, and durability of bridge structures during operation.

No MeSH data available.