| Autonomous damage detection in engineering structures is critically important as the demands placed upon those structures are increased. It has been well established that damage detection can be done through monitoring the modal frequencies of the structure for changes, and furthermore; through the use of various techniques, it is possible to relate the pattern of modal frequency changes to a specific damage state. These forward methods generally require good sensitivity of natural frequency changes to damage states as well as a rich data set to avoid ill-conditioning. In order to deal with the low sensitivity of natural frequencies to structural changes, sensitivity-enhancing controllers are used to amplify these changes. Through the use of these controllers, coupled with the open-loop structural response, it is possible to improve the results of existing damage identification methods or to implement those methods in situations where it would not have otherwise been possible due to insufficient modal data. This work extends these forward techniques through the inclusion of additional modal properties. These techniques are applied in a laboratory situation, where it becomes critical to be able to identify a numerical model of the structure from available input-output data. If it is possible to identify a structural model, this model becomes the analytic model that is used to develop damage hypothesis data. The identified model is also the source of the measured modal data used when comparing experimentally measured damage cases to analytically derived damage hypotheses. It is shown that damage identification is very possible in relatively simple structures where high fidelity structural models either exist or can be identified. It is also shown that in a more complex structure, damage identification becomes much more challenging. |
