An enhanced ground motion selection algorithm for seismic safety assessment of structures

This paper proposes a ground motion selection algorithm (GSA) incorporating a multi-objective optimization scheme for selecting site-specific ground motion records for nonlinear time history analysis. The optimization algorithm utilized is a modified NSGA-II algorithm (MNSGA-II) with a custom initialization procedure for ground motion selection that assists in finding records that efficiently match the mean and variance of conditional spectrum. The proposed GSA is applied to three case study regions and is compared with other GSAs such as the greedy optimization algorithm and the original NSGA-II algorithm to test its efficiency. The results show that the proposed GSA efficiently returns records optimally fitting the target spectrums' mean and variance. Moreover, the results show that the proposed optimization scheme converges faster and gives a more optimal results compared to other GSAs.