Generation of Near-Fault Artificial Records using Artificial Intelligence

Due to the scarcity of ground motions, it is vital to generate appropriate artificial records in order to perform nonlinear dynamic analysis, particularly in near-field regions. In this paper a novel methodology is proposed to generate pulse-like ground motions. The generation process includes simulation of nonpulse-type high frequency component of ground motions and directivity pulses separately and then combining them to accomplish final pulse-like ground motion. Neuro-fuzzy networks have been used to produce spectrum compatible nonpulse-type ground motions. A smoothening approach is taken in order to extract directivity pulses from training records. PSO is employed to train Neuro-Fuzzy networks using optimized rules and membership functions. Wavelet transform is used to decompose accelerograms to special range of frequencies. PCA is used as a dimension reduction technique in order to improve training efficiency. At the end, an example is provided to show the efficiency of the proposed method

کلیدواژه ها:

Near-fault ، Artificial record ، Neuro-Fuzzy ، Wavelet transform