PSI - Issue 64

Marco Martino Rosso et al. / Procedia Structural Integrity 64 (2024) 507–514

512

6

Author name / Structural Integrity Procedia 00 (2019) 000–000

Fig. 5. Equivalent shear-type lumped mass 3D model of existing RC frame for the numerical validation of i-AOMA.

4. i-AOMA dynamic identification results and discussion

The numerical acceleration response data have been decimated with a factor equal to 5, thus restricting the frequency realizations domain within the Nyquist frequency of 20 Hz. The singular value decomposition of the power spectral density evidenced that the collected signals are informative since their peaks evidenced the natural frequencies of the system, except for the first mode at 1.23 Hz which appeared obscured by the spectral noise (see Fig. 6). Despite the SSI-cov control parameters sampling intervals have been automatically computed by the i AOMA throughout the relationship illustrated in Rosso et al (2023), to further enhance the exploration step 1, the maximum model order has been limited to half of the theoretical possible value, i.e. to 680. Totally, 308 control parameters samplings have been carried out in the i-AOMA step 1 for collecting 200 user-defined successful SSI cov analyses (success rate of 64.9 %). Consequently, to the stability checks, the stabilization diagram obtained by overlapping all the fully stable poles has been analyzed through the FFT-KDE algorithm delivering a bandwidth parameter of 0.00217 Hz. Therefore, 11 natural frequencies out of the theoretical 12 ones, and their stable poles’ alignments, have been selected around the peaks of the normalized KDE, i.e. those peaks overcoming the statistical based prominence value of 0.093.

Fig. 6. Singular value decomposition of the power spectral density, typical of the frequency domain decomposition OMA method.