Damage Detection with Closely Spaced Modes Using Autocovariance Functions

Monitoring of structures having closely spaced modes is challenging due to issues in system identification and mode tracking. However, these issues are not present if damage detection is performed in the time domain using e.g. raw sensor data or covariance functions. Replacing raw vibration measurement data with autocovariance functions (ACFs) offers many advantages in damage detection. For stationary random excitation, ACFs have the same form as a free decay of the system. Consequently, ACFs of different sensors are spatiotemporally correlated, which can be utilized in damage detection. In case of closely spaced modes, correlation between modal responses is non-negligible, which can make damage detection more difficult. Damage detection with closely spaced modes was explored numerically using a finite element model of a frame structure subject to random excitation and variable environmental conditions. Acceleration measurements were simulated at three positions. Damage was a local stiffness degradation. Also, experimental data of a lattice tower under wind excitation were analyzed. Damage was a stiffness and mass decrease due to removal of bracings. It was found that closely spaced modes or mode crossings did not require any special treatment if damage detection was performed in the time domain.