Extraction of dynamic features from short acceleration data bursts: a review

It is well known that structural damage can lead to changes in dynamic features such as frequencies, mode shapes, damping, vibration intensity, etc. Signal processing tools available to extract these features include Wavelet analysis, Fourier and Hilbert-Huang transforms. Acceleration data is typically used as input to these tools, given that it is a type of response with a relatively high dynamic component (i.e., oscillations in the response due to inertial forces of the structure) in relation to the static component (i.e., derived from time-varying static deflections as a result of time/spatial-varying loads). Almost all traditional signal processing approaches require access to long-time data sets. For instance, long periods of acceleration and multiple measurement points allow engineers to accurately define the mode shapes of a structure. In this paper, a scenario is envisioned where drones are used to charge sensors placed on bridges as well as to acquire the data recorded by the sensors for processing. The novelty is the challenge of monitoring structural condition in the context of acquiring limited quantities of data. The latter requires being able to deal with a very significant impact of edge effects and the loss of resolution due to the short duration of the signal. This paper reviews attempts to obtain bridge dynamic features overcoming these limitations, i.e., via multi-stage measurements as in the case of the Short Time Frequency Domain Decomposition method.