Detection of structural damage through changes in frequency

Among all the structural vibration characteristics, natural frequencies are relatively simple and accurate to measure, and provide the structural global damage information. In this paper, the feasibility of using only natural frequencies to identify structural damage is exploited by adopting two usual approaches, namely, sensitivity analysis and neural networks. Some aspects of damage detection such as the problem of incomplete modal test data and robustness of detection are considered. A laboratory tested 3-storey frame is used to demonstrate the possibility of frequency-based damage detection techniques. The numerical results show that the damaged element can be correctly localized and the content of damage can be identified with relatively high degree of accuracy by using the changes in frequencies.