A fully autonomous damping estimation with SVM-based fault data treatment using 1-year wireless monitoring data

This study reports on the estimated modal damping ratio of a parallel cable-stayed bridge by the use of automated Operational Modal Analysis (OMA). The 1-year monitoring data from a dense wireless smart sensor network (WSSN) of 113 smart sensors were utilized for damping estimation. A novel data treatment strategy for sensor fault in WSSN data was proposed to remove a static trend, recover the unexpected spikes, and exclude the fault measurements autonomously. The automated covariance driven Stochastic Subspace Identification (SSI-COV) is determined as the OMA algorithm. In order to achieve more reliable damping estimates, the three-stages of validations were implemented in SSI-COV for the purpose of eliminating spurious poles from physical poles. The improvement in the integrated damping estimation procedure was demonstrated by comparative results of OMA-based damping estimation of the Jindo Bridge, by using a raw and treated data. The effect of data length on the accuracy of damping estimates was evaluated statistically.