Developing the seismic fragility analysis with fuzzy random variables using Mouth Brooding Fish algorithm

Abstract

The present work proposes a fuzzy-random fragility assessment framework for evaluating 2D reinforced concrete moment frame buildings in the presence of various sources of uncertainty, most notably aleatory and epistemic. Such uncertainties exert a powerful influence on the nonlinear behavior of structural systems and consequently affect the seismic response of these structures. For this reason, a number of effective techniques including Latin Hypercube Sampling (LHS) simulation, fuzzy set theory, and a well-known α-cut approach have been used to quantify the median of the collapse fragility curve as the fuzzy-random response. As a major step in the α-cut approach, metaheuristic evolutionary algorithms including modified genetic algorithm (MGA), and a novel global optimization algorithm inspired by Mouth Brooding Fish (MBF) in nature have been adopted to explore the maximum and minimum of such median in each membership degree, α. The results demonstrate that the merit of new MBF algorithm is its greater efficiency in specifying the α-cut boundaries compared with MGA. Herein, for the sake of more simplicity and efficiency, a new equation is proposed for the prediction of the median of collapse fragility curve of the case-study building, using the gene expression programming (GEP) methodology. This median is formulated in terms of several effective parameters such as steel modulus of elasticity Es, steel yield stress fy, and concrete strength f′c, which regarded as the input fuzzy-random variables. The performance and validity of the GEP model are further tested using several criteria. The results indicate that analyzing the proposed GEP model in terms of fuzzy-random variables thru the MBF algorithm significantly improves efficiency and reduces computational time by 75%.