Seismic reliability analysis of high-pier railway bridges with correlated random parameters via an improved maximum entropy method

Seismic reliability analysis is crucially important for the seismic performance evaluation and probabilistic seismic risk assessment of bridges. However, the exact and efficient computation of seismic reliability for a practical bridge with uncertainty is still challenging. This study presents a composite method of maximum entropy principle with fractional moments, partially stratified sampling, and Nataf transformation for the seismic reliability analysis of bridge with correlated random parameters, and then evaluates the impact of random variable correlation on the probabilistic seismic performance of high-pier continues rigid frame bridge (CRFB). Different from the existing maximum entropy method, a unbiased likelihood function-based maximum entropy method is proposed to compute the extreme value distribution (EVD) of structural nonlinear seismic response. To illustrate the efficiency and accuracy of the proposed method, a hysteretic nonlinear single-degree-of-freedom (SDOF) system and a simplified two-span bridge are first taken as examples. The obtained EVD is compared with the results of Monte Carlo simulation, current maximum entropy method, and widely used probability distribution model fitting. The proposed method is then applied to evaluate the seismic reliability of a practical high-pier CRFB. The influence of uncertainties and variable correlation of structural parameters on the seismic reliability of high-pier CRFB is discussed in detail. The analysis results indicate the high efficiency and accuracy of the proposed method and demonstrate the significant importance of structural parameter correlation on the probabilistic seismic performance assessment of high-pier CRFB.