Seismic fragility analysis is an efficient method to evaluate the structural failure probability during earthquake events. Among the existing fragility analysis methods, the probabilistic seismic demand model (PSDM) and the joint probabilistic seismic demand model (JPSDM) are generally used to compute the component and system fragility, respectively. However, the statistical significance behind the parameters related to the current PSDM and JPSDM are not comparable. Aside from that, when calculating the system fragility, the Monte Carlo sampling (MCS) method is time-consuming. To solve the two flaws, in this paper, the logarithm piecewise functions were used to generate the PSDM and the JPSDM, and the MCS was replaced by the univariate conditioning approximation (UCA) method. The concepts and application procedures of the proposed fragility analysis methods were elaborated first. Then, the UCA method was illustrated in detail. Finally, fragility curves of a steel arch truss case study bridge were generated by the proposed method. The research results indicate the following: (1) the proposed methods unify the data sources and statistical significance of the parameters used in the PSDM and the JPSDM; (2) the logarithmic piecewise function-based PSDM sensitively reflects the changing trend of the component’s demand with the fluctuation of the seismic intensity measure; (3) under transverse seismic waves, major injuries happen on the side bearings of the bridge, while slight damage may occur on each pier, and as the seismic intensity measure increases, the side bearings are more likely to be damaged; (4) for the severe damage and the absolute damage of the studied bridge, the system fragility curves are closer to the upper failure bounds; and (5) compared with the MSC method, the accuracy of the UCA method can be guaranteed with less calculation time.
Seismic Fragility Analysis of the Reinforced Concrete Continuous Bridge Piers Based on Machine Learning and Symbolic Regression Fusion Algorithms
