APPLICATION OF NEURAL NETWORK IN RELIABILITY PREDITION OF SEISMICLY ISOLATED STRUCTURES SUBJECTED TO RANDOM GROUND MOTIONS

One of the most effective technologies of seismic resistant design of structures is base isolation which has lots of different types due to their mechanical behavior. In this study Friction Pendulum Bearings (FPBs) as one of the popular types of base isolation is applied on a specified structure. Due to stochastic nature of variables such as input ground motion; a novel method is proposed to predict the reliability of the supposed structure using artificial neural networks (ANN). The reliability of the system in the format of probability of failure (Pf) is calculated using a simulation based method which is an effective tool for an isolated structure subjected to random earthquake excitations. A 2D concrete frame three-story structure isolated with FPB, representing critical facilities, such as a data center, is considered as the super structure. The super structure is designed for gravitational and lateral loads based on ACI 318-05. Random excitations are applied by the means of artificial earthquake ground motions generated through the superposition of a random ground velocity record with a single, coherent, long-period velocity pulse. The probability of failure for a particular set of structure and isolation parameters was calculated using Monte Carlo Simulation by time history structural analysis at first. Then a set of neural networks were trained to predict the peak responses of the structure. Six random parameters of artificial earthquake ground motion were assumed to be the input variables of neural networks. The probability of failure was calculated again, using neural networks. The results show a good compatibility to the ones calculated using time history structural analysis