Diagnostics of damage presence using tomographicaily constructed probability distribution based on Lamb wave signals

A correlation analysis for constructing the probability distribution of damage presence was developed, using guided Lamb wave signals from a multisensor network. Instead of using conventional parameters, such as time-of-flight (ToF), correlation coefficients deduced from the acquired Lamb wave signals retrieved for each actuator-sensor path between the reference state (benchmark) and present state (damaged) were used for estimating damage. The image of the potential damaged region was therefore constructed tomographicaily to track the damage presence and development. Experimental validation was conducted using an aluminum plate with nine surface-mounted piezoelectric elements. A Shannon entropy optimization criterion was applied to calibrate the most optimal mother wavelet and the most appropriate continuous wavelet transform scale for wave signal processing. The prediction from the algorithm agrees well with the actual damage status (both crack location and crack orientation), demonstrating the satisfactory performance of the proposed method for damage identification in thin plate-like structures.