Abstract: Empirical Mode Decomposition (EMD) is a technique that converts the
measured signal into a number of basic functions known as intrinsic mode
functions. The EMD-based damage detection algorithm relies on the principle
that a sudden loss of stiffness in a structural member will cause a
discontinuity in the measured response that can be detected through a
distinctive spike in the filtered intrinsic mode function. Recent studies have
shown that applying EMD to the acceleration response, due to the crossing of a
constant load over a beam finite element model, can be used to detect a single
damaged location. In this paper, the technique is further tested using the
response of a discretized finite element beam with multiple damaged sections
modeled as localized losses of stiffness. The ability of the algorithm to
detect more than one damaged section is analysed for a variety of scenarios
including a range of bridge lengths, speeds of the moving load and noise
levels. The use of a moving average filter on the acceleration response, prior
to applying EMD, is shown to improve the sensitivity to damage. The influence
of the number of measurement points and their distance to the damaged sections
on the accuracy of the predicted damage is also discussed.
Keywords: Empirical Mode Decomposition, vibration-based analysis, damage detection, signal decomposition
