Abstract
Guided-wave-based structural damage identification techniques have received more and more attention in the civil engineering community. They not only have the capability of detecting smaller damages on a structure than vibration-based methods, but also can cover a relatively larger magnitude, compared with other traditional non-destructive evaluation techniques. To realize damage identification, features usually need to be extracted from the time domain responses. This is achievable for homogeneous materials, including steel and aluminum. But for composite materials, such as concrete, the features are usually very difficult to be extracted, because of the existence of small aggregates and the nature of uneven material properties which generate multiple reflections. It is very difficult to simulate the time domain responses and to identify damages by using time domain responses directly for such random material. Recently, a new damage identification scheme is proposed, named as DISC (Damage Identification based on Sparse Coding). This method is essentially a pattern recognition technique, which avoids the traditional fixed transform process but takes advantage of the existing data by dictionary learning techniques. This paper will review the DISC method and then apply it to identification of de-bonding damage in concrete beam using guided wave test data. The results will demonstrate the effectiveness of the DISC methodology.