Damage Detection in Nonlinear Elastic Structures Using Individual Sensors

Natural frequencies have always been one of the most intuitive and widely used features for damage identification in civil structures. Even with the recent rapid technological and theoretical developments, frequency-based identification methods are of great interest for applications through low-cost sensing systems. Although most techniques for frequency identification assume a linear structural behavior, in real applications, variations in the amplitude of input excitation can lead to short-term frequency fluctuations due to the inherent nonlinearities of civil structures. This paper proposes a procedure for damage detection in nonlinear systems based on instantaneous resonant frequency and amplitude estimates. A statistical model was fitted to identified data, and a synthetic indicator was proposed to obtain robust damage detection, even when frequency shifts due to variations in the input excitation are comparable to those due to actual damage. The proposed method was applied to a dataset recorded from a reinforced concrete building with strongly nonlinear behavior.