Performance analysis of vibration-based damage indicators under low-modal information structures

There exists a wide range of frequency-based damage indicators, which have been verified under high-modal information structures. Nonetheless, their performance and behaviour under less advantageous conditions remain unanswered. In this work, an original methodology is presented, combining an analysis of the damage detection and a statistical method to assess the damage quantification of each indicator. Furthermore, this study is validated using experimental data from three increasingly complex structures due to the limited modal information. The approach calculates different damage features based on the vibration responses of the structures. Subsequently, the prediction obtained from damage features is compared with the actual condition of the structure. The performance of each damage indicator is quantified using the Area under the Curve. In parallel, a correlation between damage indicator response and stiffness loss is conducted to understand the damage severity quantification of indicators. In addition, the minimum detectable damages are calculated and reported for different confidence intervals. The results showed that matrix-based indicators present advantages over other documented methods. Moreover, the influence of noise and modal information was studied, demonstrating that their diminishing effects do not equally impact all damage indicators. The findings of this work give insights into how each damage indicator performs in different structures and conditions. Finally, this work proposes an objective methodology to compare damage detection strategies.