Abstract
In the structural reliability analysis of large and complex systems, it is necessary to develop a calculation method of failure probability with efficiency and accuracy in order to quantitatively evaluate the safety. So far, Monte Carlo simulation has been used because of its simplicity and easiness of calculation. For large and complex structures, many improved Monte Carlo simulation methods have been proposed, which can efficiently obtain various failure samples. In this study, an attempt is made to develop an effective sampling method for large-scale structures with multiple failure modes by using Multi-Agent Optimization (MAO) which is one of meta-heuristic optimization techniques. The proposed method is first applied to a one-bay one-story simple rigid frame structure with a limit number of failure mechanism so as to investigate the applicability of the proposed method. From the numerical results, it is obtained that the original MAO can obtain a few failure samples with high occurrence probability, but can not find other lower probability samples necessary for the reliability assessment of complex systems. Therefore, a new method called Resampling MAO is proposed in this study, which successively estimates appropriate searching spaces according to the occurrence probabilities of failure samples. Failure samples are searched in descending order of failure probability. Therefore, it is expected that the proposed method is useful to analyze important members or elements of structures or systems and improve the efficiency of Importance Sampling.
