| This thesis investigates vibration-based damage identification problems in smart structures. Vibration-based damage metrics suffer from small sensitivity of modal properties to damage, thus making it difficult to detect, locate, and determine the extent of damage using vibration-based metrics. The key framework of the study is that one can modify the dynamics of a smart structure using the Sensitivity Enhancing Control (SEC) so that damages of small magnitude, such as fatigue surface cracks, can be easily identified. In addition, a smart structure can accommodate dual functions; one for conventional vibration control, the other one for performing autonomous structural health monitoring. SEC not only offers enhanced modal frequency sensitivity toward structural damages, but also improves the result of damage localization, even with typical measurement noise thresholds present in measured signals.; In this thesis, two damage localization methods that take advantage of the SEC concept are developed. The first method employs the linear approximation, which is based on the sensitivity evaluation of an analytic model between damage variables and measured modal frequency shifts. In this method, carefully designed closed-loop systems enable a structure to yield independent and additional modal data sets. These additional closed-loop dynamics enrich the content of modal information for improved numerical conditioning of the sensitivity-based method and thus provide more precise damage identification. The second method uses the correlation between damage hypotheses from a baseline model and measured modal frequency shifts to determine the most likely damage event. Here, SEC improves signal-to-noise ratio, such that the location of damage can easily be identified from measurements of closed-loop system frequencies while damage cannot be localized using open-loop frequencies. These damage localization methods are experimentally demonstrated in a two degree-of-freedom system subject to stiffness and/or mass damage. This thesis also fuses fatigue crack nonlinearity and closed-loop pole sensitivity, which allows non-model based damage detection in structures with fatigue cracks through enhancement of the harmonics of a given excitation frequency. The results of numerical simulation and experimental evaluation show that the performance and robustness of damage identification in smart structures can be improved significantly through SEC. |
