Artificial accelerograms composition based on the CEEMD

Abstract

In this work, a new methodology for generating spectrum-compatible accelerograms is presented. The proposed methodology considers the non-stationary and non-linear characteristics of seismic signals and utilizes the Hilbert–Huang transform (HHT) to analyse them. The two reported drawbacks of HHT, i.e. the mode mixing phenomenon and the end effects issue, are resolved through the proposed methodology. More specifically, the advantages of the recently introduced complementary ensemble empirical mode decomposition (CEEMD) are exploited in order to eliminate the mode mixing phenomenon. Moreover, the application of the proposed method only to the strong motion duration of the seismic signal assists to overcome the end effects issue. Thirty natural seismic records of different characteristics, such as frequency content, amplitude and duration, are employed as initial seed signals to demonstrate the proposed method. To illustrate the effectiveness of the proposed technique, comparisons with three established methods for generating spectrum-compatible seismic accelerograms are also provided.