Multi-parameter seismic fragility and sensitivity analysis of long-span cable-stayed bridge based on multi-task lasso regression

Abstract

The accuracy and efficiency of the seismic fragility assessment of bridges encounter challenges when the span increases and the structure becomes more complex, which is significantly affected by various uncertain parameters and has to be studied. Multi-parameter seismic fragility analysis (MP-SFA) is a powerful tool that explicitly connects fragilities and multiple variables, including structural parameters and seismic intensity measures (IM). This paper presents an efficient generation method for MP-SFA based on multi-task Lasso regression and is verified using an example of a long-span cable-stayed bridge. Based on the framework of MP-SFA, the mean fragility curves of various bridge components are established, which averages the influence of all considered random parameters. A sensitivity analysis method is also established to identify significant variables for fragility function. The analysis and numerical results show the necessity of MP-SFA to overcome the dispersion of bridge fragility due to the incapability of a single parameter. PGV is shown as the most significant parameter for the case. However, the other IMs also have a non-negligible impact on fragility.