Generation of Near-Field Artificial Ground Motions Compatible with Median-Predicted Spectra Using PSO-Based Neural Network and Wavelet Analysis

The principal purpose of this article is to present a novel method based on particle swarm optimization (PSO) and wavelet packet transform (WPT) techniques and multilayer feed-forward (MLFF) neural networks, in order to generate spectrum-compatible near-field artificial earthquake accelerograms. PSO is employed to optimize the weight values of the networks. Moreover, to improve the training efficiency principal component analysis (PCA) is used, as a reduction technique. The proposed PSO-based MLFF (PSOBMLFF) neural network develops an inverse mapping from compacted and reduced spectrum coefficients into metamorphosed accelerogram wavelet packet coefficients. In this research, to produce compatible synthetic long-period near-field ground motions with median predicted spectra, the attenuation of peak ground velocity (PGV) with the close horizontal distance (R), moment magnitude (M), and time-average shear-wave velocity from the surface to 30 m (Vs30) has been developed using nonlinear regression analysis.