Hybrid two-step method of damage detection for plate-like structures

This paper develops a two-step approach for structural damage detection. In the first step, the two-dimensional generalized local entropy (2D-GLE) is proposed to identify the damage location and number for plates. Based on the 2D-GLE, singularities in plates are revealed with the aid of mode shapes; hence, damages are localized, and the number of which is estimated. Besides, the statistical analysis is applied in the analysis of 2D-GLE damage maps to overcome the difficulty in damage detection methods without baseline data. In the second step, the damage severities at the identified locations are evaluated via the optimization using artificial bee colony (ABC) with the designed frequency object function. The advantages of the present method are as follows: (1) No-baseline in the measurement of mode shapes. Damage localization via the 2D-GLE avoids using healthy mode shapes serving as baselines in damage detection, and then only damaged mode shapes are required. (2) No down-sampling. Compared with the wavelet-based methods, no down-sampling enhances the algorithmic robustness and accuracy. (3) Efficiency. This hybrid two-step method shrinks the solving parameter domain sharply in optimization via the first step. Hence, only the severity parameters need to be investigated in ABC. (4) Robustness. ABC ensures the robustness of the proposed method. Different damage cases are investigated in simulations, single-damage/multiple-damage cases, and noise immunity tests demonstrate that the proposed method is effective and accurate in damage detection. At last, a simple experiment is given to verify the present algorithm.