Modal parameter identification of a curved cable-stayed model bridge based on EDA and DATA-SSI

For bridge health monitoring, the measured data may be unreliable due to various interferences. In order to get a reliable modal parameter identification result of a bridge, it is vital to have an inspection and a pre-processing on the bridge health monitoring data. Firstly, exploratory data analysis (EDA) was adopted to inspect the data quality, and the unreliable data measured from malfunctioning sensor was removed. Then, outlier analysis was performed to eliminate the abnormal data points from the data set. In the end, data driven stochastic subspace identification (DATA-SSI) combined with stabilization diagram was applied to identify the bridge modal parameters. A large scale curved cable-stayed model bridge was taken as an instance to verify the proposed method. The comparison of the modal parameter identification results of the original and the pre-processed data shows that the proposed method is effective, accurate and valuable.‌