Operational Modal Analysis (OMA)-based damping estimation of large-scale structures is challenging issues due to its large error. Meanwhile, the rapid development of wireless sensor networks (WSN) has enabled dense instrumentation in civil infrastructure, and this dense deployment of WSN allows a more stable modal analysis. Nevertheless, WSN still exhibits difficulties in damping identification owing to the existence of the faulty data resulted from its vulnerability to harsh environments than wired monitoring systems. This study focuses on the enhanced damping estimation by the newly proposed fault-data management combined with an automated OMA algorithm. The Support Vector Machine is utilized for automated detecting, recovering, and isolating faulty data in WSN. A new feature called Maximum Correlation Factor (MCF) is proposed to measure the similarity between simultaneously measured data sets. The three features of kurtosis, Dispersion Rate, and MCF are utilized in combination for the training and validation process with two different kernel functions. A fully automated OMA algorithm of Covariance driven Stochastic Subspace Identification is utilized to estimate a damping ratio from the output-only measurement on operational condition with three-stage validations for eliminating spurious poles. The accuracy and reliability of the proposed damping estimation algorithm are demonstrated by a 9-DOF numerical simulation and to the 1-year WSN monitoring data collected from the Jindo Bridge.