Abstract:
Bridges are subjected to a variety of loads throughout their service life, including
service loads, seismic events, and environmental forces, which may compromise their
structural integrity. As these structures are significant and expensive assets, there is a
growing emphasis on using advanced Structural Health Monitoring (SHM) techniques
to ensure their safety and longevity. This research introduces a refined approach for the
precise detection of damage in bridges using state-of-the-art signal processing methods.
In the initial phase, ambient vibration data are captured via accelerometers strategically
installed on the bridge to record its response in healthy state for baseline data.
Vibrations data are then processed using Complete Ensemble Empirical Mode
Decomposition with Adaptive Noise (CEEMDAN) to extract intrinsic mode functions
(IMFs). These IMFs are used for signal reconstruction followed by further analysis
thorough Empirical Wavelet Transform (EWT). This combination efficiently
segregates the bridge vibrations into distinct modes, thereby enhancing the
identification of potential damages by mitigating noise interference and closely spaced
frequencies. The Spectral Centroids (SC) of these modes are computed using Short
Time Fourier Transform (STFT) to serve as critical damage-sensitive features.
Subsequently, an artificial intelligence model utilizing Kernel Principal Component
Analysis (KPCA) is employed. This model is trained on baseline data to establish
control limits and then used to detect deviations in new datasets by analyzing changes
in the Squared Prediction Error (SPE) and Hotelling’s T2 statistic, it enables effective
damage detection. The efficacy of this methodology is empirically validated using real
world data collected from the Z-24 bridge, demonstrating its practical applicability and
reliability. Additionally, an RC-bridge health condition was also monitored to detect
any potential damage. The innovative integration of CEEMDAN, EWT, and KPCA in
x
this research offers a robust framework for early damage detection, promising
significant improvements in the maintenance and monitoring of bridge structures, the
efficiency of which was validated through performance metrics.
