STOCHASTIC GENERATION OF ARTIFICIAL ACCELEROGRAMS USING THE CONTINUOUS WAVELET TRANSFORM METHOD Hera Yanni 1 , Michalis Fragiadakis 1 , and Ioannis P. Mitseas 1,2 1 National Technical University of Athens, Athens, Greece Iroon Politechniou 9, 15780, Athens, Greece e-mail: heragian@mail.ntua.gr; mfrag@mail.ntua.gr; Imitseas@mail.ntua.gr 2 University of Leeds, Leeds, United Kingdom LS2 9JT, Leeds, UK e-mail: I.Mitseas@leeds.ac.uk Keywords: artificial a ccelerogram, s tochastic m odel, t emporal a nd s pectral non-stationarity, continuous wavelet transform, wavelets, spectrum-compatible. Abstract. A new stochastic methodology for the generation of artificial, fully non-stationary, spectrum-compatible accelerograms is proposed. Time-history analysis requires the use of suites of accelerograms that can be either recorded, or artificially generated. Hazard con- sistency can be achieved if their mean spectrum is matched to a target design spectrum. Ar- tificial earthquake ground motions can be easily generated at frequency bands of interest and have features compatible with desired target characteristics. The proposed model operates in the time-frequency (TF) domain using the continuous wavelet transform (CWT). The method- ology follows the rationale of the techniques that use an evolutionary power spectral density function, where the temporal and spectral non-stationarity is modeled by a TF envelope func- tion. More specifically, a zero mean stationary Gaussian stochastic process is generated using the spectral representation method and defines: the target peak ground acceleration, the fre- quency range, and the phase distribution. Additionally, it ensures the spectrum compatibility through the power spec-tral density function. A recorded accelerogram is consistent with the characteristics of the site of interest and models the temporal and spectral modulation. Both generated and recorded signals are analyzed in the TF domain using the CWT and their wavelet coefficients are obtained respectively. The time-frequency envelope is then extracted from the modulus of the CWT coefficients of the recorded accelerogram and modifies accordingly the stationary stochastic pro-cess in the TF domain. Energy compatibility is also imposed at each frequency, and new wavelet coefficients are produced. The simulated signal is then transformed in the time domain using the inverse CWT. The proposed methodology provides an arbitrary number of seismic accelerograms whose temporal and spectral modulation is modeled by a recorded ground motion. 5143 COMPDYN 2023 9 th ECCOMAS Thematic Conference on Computational Methods in Structural Dynamics and Earthquake Engineering M. Papadrakakis, M. Fragiadakis (eds.) Athens, Greece, 12-14 June 2023 Available online at www.eccomasproceedia.org Eccomas Proceedia COMPDYN (2023) 5143-5154 ISSN:2623-3347 © 2023 The Authors. Published by Eccomas Proceedia. Peer-review under responsibility of the organizing committee of COMPDYN 2023. doi: 10.7712/120123.10790.22279