摘要

This paper reports a two-step approach for detecting multiple damages in thin plates. The first step focuses on damage location detection. This is done by applying the 2-D wavelet transform to the modal shape to reveal singularities and hence the damage locations. The damage severities at the identified locations are then assessed in the second step using a particle swarm optimization (PSO) algorithm. Both the modal shape used in the first step and the severity evaluation database required for the second step to reveal the relationships between natural frequencies and damage severities are obtained using a wavelet finite element method. Our simulation studies of a simply supported plate with multiple damages demonstrated that the proposed method is effective in multi-damage detection and performed reasonably well even when the natural frequencies are not precisely measured. Our analyses also suggest that the use of higher natural frequencies could lead to better results for damage severity evaluation. We also find that it is desirable to include more natural frequencies in order to get more accurate severity evaluation results.