New generalized ANN-based hybrid broadband response spectra generator using physics-based simulations

Abstract

To evaluate the seismic risk associated with infrastructures, site-specific seismic hazard studies are needed to be performed. Further, for nonlinear time history analysis, one requires broadband ground motion. In modern times using physics-based simulations (PBS) for deriving the ground motion for future earthquakes has got due consideration. The PBS helps in decreasing the uncertainties related to the hazard estimation compared to GMPEs. There is a certain threshold to the PBS methods with the limited computational facilities. Hence, certain hybrid methods are required to attain a Broadband spectra for the simulated ground motion. This study uses a new ANN-based model to generate broadband ground motion spectra using the low-frequency spectral acceleration from PBS, source, path, and site parameters as input variables.A detailed parametric study and performance evaluation was made to 1 Springer Nature 2021 L A T E X template 2 ANNHBB identify the optimal input parameters in conjunction with the best-suited ANN architecture. The performance of the ANN model is demonstrated for Iwate (Mw 6.9,2008) earthquake. We found that the predicted values from the developed ANN model are in good agreement with the recorded data. Furthermore, time histories are generated using the spectral ordinate matching technique from the estimated broadband spectra.

ANN-Based Ground Motion and Physics-Based Broadband Models for Vertical Spectra

“…The choice of the number of hidden layers and neurons is similar to GMM. Additionally, the transfer functions between inputhidden and hidden-output neurons are hyperbolic tangent and pure-linear, consistent with Sharma et al (2023). The final functional relationship between input and output variables is expressed as follows:…”

“…The model leverages low-frequency spectra obtained from physics-based simulations, which are often constrained by computational limitations in simulating high frequencies. The corresponding architecture has proved to be more efficient than the conventional BB models in developing spectra without jumping at the crossover period and in quantifying the spatial coherency characteristics appropriately (Infantino et al, 2021;Sharma et al, 2023). Hence, we intend to develop an ANN-based broadband ground motion model for vertical spectra.…”

“…Here, the ANN model is developed to obtain broadband spectra from long-period spectral acceleration (Sa) values and source (M w , F ), distance (R rup ) and site (V s30 ) parameters. The inclusion of source, path and site characteristics alongside low-frequency spectra to the model is adopted based on the advantages in predictions and performances as discussed in Sharma et al (2023). The choice of the number of hidden layers and neurons is similar to GMM.…”

“…In the study, models were provided for both Rotd50 and vertical spectral acceleration (S a ) values developed using the SIMBAD database, which consists of 500 strong motion records. Sharma et al (2023) extended the ANN model by using a more robust NGA-West2 database and including the source and site terms as the predictor variables.…”

“…Furthermore, we endeavour to develop a new broadband model for vertical spectra, recognizing its significance in non-ergodic ground motion simulation and hazard assessment. The functional form of the developed BB spectra model is similar to Sharma et al (2023). The developed model effectively captures the physical trends inherent in earthquake records.…”